Australian Government Plans New Media Censorship Laws

A report to government on the adequacy of
digital platforms’ disinformation and news

quality measures
JUNE 2021
Canberra
Red Building
Benjamin Offices
Chan Street
Belconnen ACT
PO Box 78
Belconnen ACT 2616
T +61 2 6219 5555

F +61 2 6219 5353
Melbourne
Level 32
Melbourne Central Tower
360 Elizabeth Street
Melbourne VIC
Gure 15
PO Box 13112
Law Courts
Melbourne VIC 8010
T +61 3 9963 6800
F +61 3 9963 6899
Sydney
Level 5
The Bay Centre
65 Pirrama Road
Pyrmont NSW
PO Box Q500
Queen Victoria Building
NSW 1230
T +61 2 9334 7700
F +61 2 9334 7799
Copyright notice
https://creativecommons.org/licenses/by/4.0/
With the exception of coats of arms, logos, emblems, images, other third-party material or devices protected by a trademark, this content is made available
under the terms of the Creative Commons Attribution 4.0 International (CC BY 4.0) licence.
We request attribution as © Commonwealth of Australia (Australian Communications and Media Authority) 2021.
All other rights are reserved.

The Australian Communications and Media Authority has undertaken reasonable enquiries to identify material owned by third parties and secure permission
for its reproduction. Permission may need to be obtained from third parties to re-use their material.
Written enquiries may be sent to:

Manager, Editorial Services
PO Box 13112
Law Courts
Melbourne VIC 8010
Email: info@acma.gov.au
Contents
Executive summary 1
1. Introduction 6
2. Environmental assessment 8
3. Code development 39
4. Assessment of the code 48
5. Assessment of platform performance 64
6. Considerations for future reform 76
Appendix A: Full list of recommendations and findings 85
Appendix B: Signatory assessment reports 90
Appendix C: Timeline of key events 98
Appendix D: International regulatory approaches 123
Appendix E: Other Australian Government initiatives 134
Appendix F: Development of key performance indicators 140
acma | iii
Executive summary
In December 2019, as part of its response to the Australian Consumer and
Competition Commission’s Digital Platforms Inquiry, the Australian Government
requested that digital platforms in Australia develop a voluntary code of practice to
address online disinformation and news quality.
The Australian Code of Practice on Disinformation and Misinformation 1 (the code) was
launched by industry association Digital Industry Group Inc (DIGI) on 22 February
2021. The code has since been adopted by 8 digital platforms – Google, Facebook,
Microsoft, Twitter, TikTok, Redbubble, Apple and Adobe.
The ACMA was tasked with overseeing the development of the code and reporting to
the government on the adequacy of platform measures and the broader impacts of
disinformation in Australia. Our report provides new consumer research on users’
experience of disinformation and misinformation on digital platforms and our
assessment of the industry’s code. It also provides a range of findings and a number
of recommendations for consideration by the government.
The online propagation of disinformation and misinformation presents an

increasing threat to Australians
Over the previous 18 months, we have seen increasing concern within the community
over the ‘infodemic’ of online disinformation and misinformation, particularly in relation
to the real-world impacts of COVID-19. The propagation of these falsehoods and
conspiracies undermines public health efforts, causes harm to individuals, businesses
and democratic institutions, and in some cases, incites individuals to carry out acts of
violence.
To understand the scale and impacts of this issue in Australia, we undertook a mixed-
methods study focused on COVID-19 misinformation. Key insights include:
> Most adult Australians (82%) report having experienced misinformation about
COVID-19 over the past 18 months. Of these, 22% of Australians report
experiencing ‘a lot’ or ‘a great deal’ of misinformation online.
> Belief in COVID-19 falsehoods or unproven claims appears to be related to high
exposure to online misinformation and a lack of trust in news outlets or
authoritative sources. Younger Australians are most at risk from misinformation,
however there is also evidence of susceptibility among other vulnerable groups in
Australian society.
> Australians are most likely to see misinformation on larger digital platforms, like
Facebook and Twitter. However, smaller private messaging apps and alternative
social media services are also increasingly used to spread misinformation or
conspiracies due to their less restrictive content moderation policies.
> Misinformation typically spreads via highly emotive and engaging posts within small
online conspiracy groups. These narratives are then amplified by international
influencers, local public figures, and by coverage in the media. There is also some
evidence of inorganic engagement and amplification, suggesting the presence of
disinformation campaigns targeting Australians.
> Many Australians are aware of platform measures to remove or label offending
content but remain sceptical of platform motives and moderation decisions. There
is widespread belief that addressing misinformation requires all parties –
1
DIGI, Australian Code of Practice on Disinformation and Misinformation, February 2021.
acma | 1
individuals, platforms and governments – to take greater responsibility to improve
the online information environment and reduce potential harms.
Digital platforms have introduced a range of measures in response to the

growth of disinformation and misinformation on their services
In response largely to global concerns, digital platforms have introduced measures
typically based on company-wide policies including:
> supporting third-party fact-checking organisations
> proactively updating their policies to specifically address unique events, such as
the COVID-19 pandemic and the 2020 US presidential election
> investing in means to signal credible, relevant and authentic information
> providing financial assistance and grants to news outlets, government and not-for-
profit organisations to bolster the spread of credible information and news
> increased detection, monitoring and enforcement action against groups and
networks who use their services to spread disinformation and misinformation.
Despite platforms’ mostly global approach to updating policies and implementing other
actions, many measures have had an impact on Australian users.
> In 2020, Facebook removed more than 110,000 pieces of COVID-related
misinformation generated by Australian accounts.
> Between July and December 2020, Twitter removed 50 pieces of content authored
by Australian accounts for contravening its COVID-19 misleading information
policy.
> In 2020, Google blocked 101 million advertisements globally for contravening its
misrepresentation policies.
> TikTok’s COVID-19 Information Hub was visited by over 292,000 Australians
between November 2020 and March 2021.
The above data shows that platforms are taking proactive steps to tackle
disinformation and misinformation on their products and services. The introduction of
an Australian industry code builds on these actions to codify actions, improve
transparency, enhance consumer protections, and implement mechanisms to monitor
their effectiveness. It also provides a framework to promote stakeholder collaboration
and incentivise further actions by platforms to respond to a rapidly evolving online
environment.
Digital platforms have come together to develop a single outcomes-based code

of practice with several important features
It is extremely positive to see industry, steered by DIGI, come together to develop a
single code of practice. A single code should promote a consistent approach by
platforms and provide confidence in industry to manage the range of harms associated
with disinformation and misinformation.
DIGI ran a meaningful public consultation process in developing its draft code, which
attracted a variety of submissions that clearly influenced subsequent changes. In
particular, the scope of the code was expanded to cover misinformation as well as
disinformation, a key piece of stakeholder feedback during the consultation process.
The ACMA considers this is an improvement on the EU Code of Practice on
Disinformation.
The code adopts an outcomes-based regulatory approach that allows a range of

platforms with different services and business models to sign up to the single code.
Signatories are required to sign up to the objective of ‘providing safeguards against
2 | acma
harms that may arise from disinformation and misinformation’ and may opt-in to other
code objectives, such as disrupting advertising incentives and supporting strategic
research.
The code also provides signatories flexibility to implement measures to counter

disinformation and misinformation in proportion to the risk of potential harm.
Signatories must also report annually on the range of measures they will implement to
achieve the objectives and outcomes.
Importantly, the code also stresses the need to balance interventions with the need to
protect users’ freedom of expression, privacy, and other rights.
Our assessment identifies further improvements that should be made to the

code’s scope and the clarity of commitments
The ACMA has assessed the code to consider whether it has met the expectations set
out by the government and has identified a range of improvements.
In our view, the scope of the code is limited by its definitions. In particular, a threshold
of both ‘serious’ and 'imminent’ harm must be reached before action is required under
the code. The effect of this is that signatories could comply with the code without
having to take any action on the type of information which can, over time, contribute to
a range of chronic harms, such as reductions in community cohesion and a lessening
of trust in public institutions.
The code should also be strengthened through an opt-out rather than opt-in model.
Signatories should only be permitted to opt out of outcomes where that outcome is not
relevant to their service and be required to provide justification for the decision.
The code is also limited in the types of services and products it covers. Private
messaging is excluded, despite increasing concern about the propagation of
disinformation and misinformation through these services, particularly when used to
broadcast to large groups. Including messaging services within the code, with
appropriate caveats to protect user privacy (including the content of private
messages), would provide important consumer protections.
We also consider improvements to the code should be made in relation to:

> its application to news aggregation services
> the treatment of professional news content and paid and sponsored content
> the weight given to news quality as a key aspect of the government’s request to
industry.
The ACMA is also concerned that the code does not place an obligation on individual
signatories to have robust internal complaints processes. This was an area of
particular concern identified in the Digital Platforms Inquiry.
The code includes commitments to establish administrative functions within 6 months

of code commencement. As code administrator, DIGI will establish a compliance sub-
committee, a detailed reporting guideline and a facility to address signatory non-
compliance. However, these functions remain under development at the time of
finalising this report. As a result, the ACMA has not been able to assess their
effectiveness.
DIGI and code signatories should consider changes to the code to address the
matters identified by the ACMA in its review in February 2022.
acma | 3
A clear and transparent measurement framework is critical to the effectiveness
of a voluntary, outcomes-based regulatory model
Signatories were required to nominate their code commitments and deliver an initial
report under the code, providing information and data on the measures they have
adopted under the code.
Signatories’ reports provide a large range of information on the actions they have
taken to address disinformation, misinformation and news quality, and their
investments in collaborative initiatives.
However, reports are heavily focused on platform outputs and lack systematic data or
key performance indicators (KPIs) that would establish a baseline and enable the
tracking of platform and industry performance against code outcomes over time.
Reports also show inconsistencies in the interpretations of key code terms and in
reporting formats.
Platforms should move quickly to identify KPIs specific to their services and work
together to establish industry-wide KPIs to demonstrate the effectiveness of the code
as an industry-wide initiative.
The ACMA recommends a number of actions by government to bolster industry

self-regulatory arrangements
The ACMA considers that it is still too early to draw concrete conclusions on the
overall impact or effectiveness of the code. The code administration framework –
including a detailed reporting guideline and mechanism to handle complaints – is not
due for completion until late August 2021. The design and implementation of these
elements will be key to the overall effectiveness of the code.
Given these circumstances, continued monitoring is required and the ACMA

recommends it provide government with another report on the code by the end of the
2022–23 financial year. This will provide sufficient time to assess the operation of the
code administration framework and assess the impact of any changes arising from the
February 2022 review of the code. As part of this report, the ACMA recommends it
continues to undertake focused research on these issues.
Initial signatory reports identify challenges in obtaining relevant data on platform

actions in Australia. Providing the ACMA with formal information-gathering powers
(including powers to make record-keeping rules) would incentivise greater platform
transparency and improve access to Australia-specific data on the effectiveness of
measures to address disinformation and misinformation. Information collected could
also be used to identify systemic issues across the digital platform industry and inform
future ACMA research.
More formal regulatory options could be considered, particularly for platforms that
choose not to participate in the code or reject the emerging consensus on the need to
address disinformation and misinformation. The ACMA recommends that government
provides the ACMA with reserve regulatory powers in relation to digital platforms –
such as code registration powers and the ability to set standards. This would provide
the government with the option to act quickly to address potential harms if platform
responses are not adequate or timely.
There are also opportunities for improved collaboration between government

agencies, platforms, researchers and non-government organisations on issues relating
to disinformation and misinformation. The ACMA recommends that the government
should consider establishing a Misinformation and Disinformation Action Group to
4 | acma
provide a mechanism to support future information sharing, cooperation and
collaboration.
The ACMA makes 5 recommendations to the government in its report.
Recommendation 1: The government should encourage DIGI to consider the findings

in this report when reviewing the code in February 2022.
Recommendation 2: The ACMA will continue to oversee the operation of the code
and should report to government on its effectiveness no later than the end of the 2022-
23 financial year. The ACMA should also continue to undertake relevant research to
inform government on the state of disinformation and misinformation in Australia.
Recommendation 3: To incentivise greater transparency, the ACMA should be

provided with formal information-gathering powers (including powers to make record
keeping rules) to oversee digital platforms, including the ability to request Australia-
specific data on the effectiveness of measures to address disinformation and
misinformation.
Recommendation 4: The government should provide the ACMA with reserve powers
to register industry codes, enforce industry code compliance, and make standards
relating to the activities of digital platforms’ corporations. These powers would provide
a mechanism for further intervention if code administration arrangements prove
inadequate, or the voluntary industry code fails.
Recommendation 5: In addition to existing monitoring capabilities, the government

support collaboration and information-sharing between digital platforms, government
agencies, researchers and NGOs on issues relating to disinformation and
misinformation.
acma | 5
1. Introduction
The Australian Consumer and Competition Commission (ACCC)’s Digital Platforms
Inquiry (DPI) found that Australians who use digital platforms to access news and
information are at risk of exposure to disinformation and misinformation.
In its December 2019 response, the Australian Government requested major digital
platforms in Australia to develop a voluntary code (or codes) of conduct for
disinformation and news quality. 2
The ACMA was tasked with overseeing the development of the code(s) and to report
to government on the adequacy of platforms’ measures and the broader impacts on
disinformation, with the first report due no later than June 2021. The government
noted that, should the actions and responses of the platforms be found not to
sufficiently respond to the concerns identified by the ACCC, it would consider the need
for any further reform.
In June 2020, the ACMA produced a position paper to guide the digital industry with its
code development. 3 The Digital Industry Group Inc (DIGI) released a draft code in
October 2020 for a 5-week public consultation. In tandem with the draft code, DIGI
also hosted a roundtable discussion with targeted stakeholders to discuss the code
contents and next steps.
A final code was released in February 2021, along with all 17 stakeholder submissions
and a summary report. 4 Initial code signatories were Google, Facebook, Twitter,
Microsoft, Redbubble and TikTok. In May 2021, DIGI announced that Apple and
Adobe had also signed up to the code and published the initial transparency reports
from all 8 signatories on its website.
Alongside the code development process, agencies across the Australian Government
have been working with platforms, impacted stakeholders and international
counterparts on a range of related initiatives to address disinformation and
misinformation (Appendix E).
Objectives of report
This report provides an examination of:
> the broader impacts of disinformation and misinformation on digital platforms –
with a specific focus on impacts in Australia (Chapter 2)
> the process to develop the Australian Code of Practice on Disinformation and
Misinformation (Chapter 3)
> the framework and content of the code (Chapter 4)
> platforms’ measures both under the code and in addition to the code (Chapter 5)
> considerations for further reform (Chapter 6).
2
Australian Government, Government Response and Implementation Roadmap for the Digital Platforms
Inquiry, December 2019, pp. 6–7.
3
ACMA, Misinformation and news quality on digital platforms in Australia: A position paper to guide code
development, June 2020.
4
DIGI, Australian Code of Practice on Disinformation and Misinformation, February 2021.
6 | acma
Methodology
The ACMA’s assessment has been informed by:
> engagement with DIGI and major platforms during the development and operation
of the code
> discussions with industry, academics, civil society, government agencies and
international regulators
> analysis of platform transparency reports
> targeted consumer research and content analysis
> ongoing monitoring and desktop research.
In this report, as guided by the code, we use the expression ‘disinformation and
misinformation’ in the broadest sense to refer to the general problem of the
dissemination of false, misleading and deceptive information online. Where a narrower
or different meaning is intended, for example, in the discussion of academic or
consumer research, this is made clear through qualifying remarks.
Defining the issue
Online disinformation and misinformation are relatively novel and dynamic

phenomena and there is no established consensus on the definition of
either term.
At a high level, many distinguish disinformation from misinformation on the

basis of intention or behaviour. On this view, disinformation is
characterised as false or misleading information created or spread to
cause harm or to deceive, and misinformation as false or misleading
information spread without an intention to harm or deceive. Some digital
platforms treat these phenomena differently; others make no distinction.
From a regulatory perspective, the ACMA recommended in its position

paper that industry take a broad view of the issue and implement a code
that addresses all kinds of false, misleading or deceptive online
information with the potential to cause harm. The ACMA suggested that
the term misinformation more accurately reflects the full scope of the issue
and the potential for harm to Australians and broader society. In the code,
DIGI has chosen to maintain a distinction between disinformation and
misinformation. Ultimately, however, it is not the terms used but the scope
addressed by the code that is critical.
The ACMA thanks representatives of the platforms, DIGI, broader industry,

academics, government agencies, impacted stakeholders and fellow international
regulators for their contribution to this assessment.
acma | 7
2. Environmental assessment
Over the last 18 months, disinformation and misinformation has become an
increasingly overt threat to Australia, and of growing concern to nearly all Australians. 5
The COVID-19 pandemic has proven to be a lightning rod for misinformation, with
governments and health officials both here and around the world recognising the
urgent and ongoing need to address the ‘infodemic’ and mitigate against its real-world
harms.
Through an examination of false, misleading or unproven narratives arising from the

COVID-19 pandemic, this chapter provides insights into the range, spread and impact
of disinformation and misinformation in Australia, providing a baseline to inform future
thinking and developments across government and industry. These findings draw on
desktop research, commissioned quantitative and qualitative research from the
University of Canberra’s News and Media Research Centre (N&MRC), 6 and a
commissioned network analysis project from creative agency and social media
consultancy We Are Social. 7
2.1. Exposure and susceptibility

In April 2020, researchers from the N&MRC asked a representative sample of
Australians about their consumption of, and engagement with, COVID-19 news,
information and misinformation (wave 1). The ACMA commissioned the N&MRC to
repeat and expand upon this research in late 2020/early 2021, undertaking a second
survey (wave 2) as well as a series of focus groups.
Most Australians experienced misinformation about COVID-19 in 2020. When asked

about the frequency of seeing news or information about the pandemic that they know
or suspect to be false or misleading, 60% of Australians reported having seen some
(low experience), and an additional 22% reported seeing ‘a lot’ or ‘a great deal’ of
misinformation (high experience). Only 7% of respondents reported no experience of
COVID-19 misinformation at all. 8
Experiences of misinformation
Most Australians have some levels of experience of misinformation. However, those
who are ‘heavy’ users of digital platforms (use of 6 or more platforms in a week) were
more likely to report experiencing high levels of misinformation (‘a lot’ or ‘a great deal’).
5
Mannheim, M. Australia Talks data shows we don't trust Instagram influencers, but advertisers rely on
them increasingly, ABC News, 27 May 2021.
6
The quantitative study consisted of a nation-wide representative survey of 2,659 adult Australians,
undertaken between 19 December 2020 and 18 January 2021, based on an earlier survey undertaken by
the N&MRC in April 2020. The qualitative study consisted of a series of 12 focus groups with a total of 60
participants, undertaken across February and March 2021. Participants were recruited based on a mix of
demographic characteristics, geographic locations, and media habits, with a particular focus on groups that
may have experiences with online misinformation but could be difficult to reach through survey research.
7
The network analysis project sought to examine the scale and drivers of 4 distinct online misinformation
narratives (anti-vaccine, anti-5G, anti-lockdown and QAnon) in Australia over a 12-month period. This
consisted of an examination of over 60,000 public conversations across Facebook, Instagram, Twitter,
YouTube and Reddit, identification and analysis of 291 Australian conspiracy-driven pages and groups on
Facebook and Instagram, and a manual review of misinformation narratives on TikTok and Telegram.
8
N&MRC, COVID-19: Australian News & Misinformation Longitudinal Study, 2021 [unpublished].
8 | acma
Figure 1: Experience of COVID-19 misinformation, by digital platform usage (%)
Heavy (6+) 29 64 3 4
Moderate (3-5) 21 62 6 10
Light (1-2) 22 54 9 14
High Low None Don't Know
Source: N&MRC, COVID-19: Australian News & Misinformation Longitudinal Study, 2021 [unpublished].
Note: High experience refers to respondents who have seen ‘a lot’ or ‘a great deal’ of misinformation. Low
experience refers to respondents who have seen ‘some’ or ‘not so much’.
Similarly, those who rely on social media as their main source of news also reported
higher levels of exposure to COVID-19 misinformation than the general population
(Figure 2). This was almost double the rate of those who rely on TV as their main
source of news (29% and 16%, respectively).
Figure 2: Experience of COVID-19 misinformation, by main source of news (%)
Social media 29 61 4 6
Online news 27 59 6 8
Radio 23 64 6 7
Print 22 55 8 15
TV 16 62 10 13
This research relies on respondents both knowing and accurately self-reporting on

their level of exposure to misinformation. To help address this limitation, the N&MRC
also asked surveyed Australians to respond to 5 claims about COVID-19 guidelines,
prevention strategies and treatments (for example, ‘wearing a mask does not
significantly reduce your risk of infection or spreading the virus’). Those who agreed
with official advice at the time for all 5 statements were considered ‘informed’ (59%),
while those who disagreed with 1 to 2 statements were considered ‘misinformed (low)’
acma | 9
(30%), and those who disagreed with 3 to 5 statements were considered ‘misinformed
(high)’ (11%). 9
Those in the ‘misinformed (high)’ category were more than 4 times as likely as the
‘informed’ respondents to report seeing a high level of misinformation (Figure 3). As
this group both report seeing more misinformation and are more likely to hold counter
views to official advice, there appears to be an association between exposure to - and
a potential belief in - COVID-19 falsehoods.
Figure 3: Experience of misinformation, by misinformed groups (%)
Misinformed (high) 57 35 3 5
Misinformed (low) 28 59 5 8
Informed 13 65 9 13
It is difficult to know whether those who are highly misinformed are actually seeing
more misinformation online, or if they are simply identifying any information that runs
counter to their worldview as misinformation. In practice, it may be a little of both. The
Pew Research Center conducted a comparable study in the United States in 2020 and
found that those who mainly get their news from social media are less informed about
major political events and more likely to have heard unproven claims about COVID-19,
like there being a connection between 5G and the virus. 10 These results suggest there
is, at a minimum, a link between reliance on social media and exposure to COVID-19
misinformation.
Given the popularity of social media as a news source in Australia, these findings may
be of some concern. The recently published 2021 Digital News Report: Australia found
that around half of the Australian adult population continue to access news on social
media on a regular basis. A further 23% of Australians nominated social media as their
main source of news – a figure that has been steadily rising each year. 11
Additionally, throughout 2020, many Australians reported relying on social media as a

key source of news and information about COVID-19. In both the April (wave 1) and
December (wave 2) surveys, social media ranked second only to traditional news
media, and was a more popular source of news and information than government
websites, health experts, and friends and family.
9
10
Mitchell, A. et al, Americans Who Mainly Get Their News on Social Media Are Less Engaged, Less
Knowledgeable, Pew Research Center, July 2020.
11
Growing from 18% in 2019 and 21% in 2020; Park, S., Fisher, C., Lee, J., K. & McCallum, K., Digital News
Report: Australia 2021, News & Media Research Centre, June 2021.
10 | acma
Figure 4: Sources of news and information about COVID-19 (%)
61
News media
63
38
Social media
34
32
Department of Health websites
27
25
Friends and family
20
20
Scientists, doctors or health experts
12
Wave 1 (April 2020) Wave 2 (Dec 2020/Jan 2021)
Notably, higher use or reliance on particular sources of news and information about
COVID-19 does not appear to equate to higher levels of trust. Australians were most
trusting of scientists, doctors or health experts (80%), and – despite its popularity –
were least trusting of COVID-19 news and information found on social media (22%). 12
A lack of trust in authoritative sources does, however, appear to be an indicator of

whether or not someone believes in COVID-19 misinformation. Those who were
‘misinformed (high)’ about COVID-19 had much lower levels of trust in scientists and
health professionals, and much greater trust in generally less reputable health and
lifestyle websites and blogs, and news found on social media (Figure 5). This accords
with other recent Australian research that shows belief in COVID-19 misinformation is
closely associated with lower institutional trust. 13
12
13
Pickles, K. et al, ‘COVID-19 Misinformation Trends in Australia: Prospective Longitudinal National Survey’,
23(1) Journal of Medical Internet Research, 2021.
acma | 11
Figure 5: Trust in news and information source, by informed and misinformed
(high) groups
11 19 16
33 29 28
36
43
23
35
21 25
20
46
87
41
58
49 46 47 44
24
16
informed misinformed informed misinformed informed misinformed informed misinformed

(high) (high) (high) (high)
Scientists, doctors, or News organisations Health and lifestyle News on social media
health experts websites and blogs
Trust Neither Distrust
Note: Misinformed category on this chart refers only to ‘misinformed (high)’ group.
Demographic differences
Those with the highest levels of experience with misinformation are more likely to be
male (28%), be aged either under 25 (38%) or between 25–40 (29%) and have a high
level of education (27%). 14 Those who fall within this demographic profile were also
much more likely to be ‘very’ or ‘extremely’ concerned about misinformation when
compared to the general population.
Age appears to be one of the strongest indicators of both exposure to, and belief in,
misinformation. Adult Gen Z respondents (born 1997 to 2003) were more than 3 times
more likely than baby boomers (born 1946 to 1964) to rely on social media for news
and information about COVID-19, as well as being 3 times more likely to fall within the
‘highly misinformed’ category.
In follow-up focus group discussions undertaken by the N&MRC, several parents,

teachers and some Gen Z participants themselves expressed concern about younger
people’s overall reliance on social media, levels of exposure to misinformation, and
their apparent willingness to accept conspiracy theories at face value.
[…] we're now trying to educate students who are getting educated
from TikTok and openly challenge you. And it's really difficult. It's very
worrying and concerning because you know that there are those kids, you'll
never make a dent in them. And it wouldn't concern me if they were just
ignorant. But it concerns me because they're being deliberately misinformed
and are holding on to that really strongly, just like any adult who believed in
QAnon.
Female, 40s, focus group #7
I think the issue is that the younger generation is a lot more impressionable
and we have access to way more misinformation than probably […] my father,
my family, of that generation. They rely a lot more heavily on credible sources
of information. […] I feel like a lot of people would just rely on social media at
14
12 | acma
our age, and that's often when, as we've mentioned before with the Facebook
algorithms and the YouTube algorithms and everything like that, it just starts
to propagate falsehoods, and I think we are probably the most vulnerable. Our
generation.
Male, 20s, focus group #10
While the quantitative data seems to suggest that exposure and belief in
misinformation is primarily an issue for younger, digitally active Australians, this is
likely not a complete picture. Both the qualitative study and our desktop research
suggest there are other, potentially more vulnerable, groups in society that could be
exposed to equally high levels of misinformation without their knowledge.
Older Australians, for example, were more likely to report in the N&MRC survey that
they did not know whether they had come across misinformation. Although this group
is generally more likely to rely on traditional sources of news, like TV and print media,
some focus group participants raised related concerns about their parents’ changing
news consumption habits, poor digital media literacy and lack of awareness of fake
news when online.
I think, like, the older generations aren't educated in this, really. And so they
are going down this rabbit hole […] I’m just thinking of my parents. [Group
agreement] […] Some of the things my Dad says sometimes, I’m like ‘Dad
that’s not right - that’s not the facts, right?’ […] he also spends time on random
sites online as well […] Whereas back in the day he always tuned into SBS
News or ABC News and that was the source of truth whereas all of a sudden
there’s all these publications and access to a lot of information.
Some culturally and linguistically diverse (CALD) communities may also be at risk of
higher exposure to online misinformation, particularly among non-English speakers.
While non-English speakers were not included in our consumer research, desktop
research suggests that immigrant communities rely much less heavily on mainstream
media sources for news and information, and more on digital media, including foreign
social media platforms and private messaging apps. 15 As noted by one focus group
participant, who identified as culturally Chinese, this can increase the likelihood of
coming across a range of falsehoods.
So [Weibo is] kind of like Twitter, but they also can do videos and lots of news,
like actual news, and some about celebrities or those kinds of gossip or
whatever […] And it is a very messy place, especially in [the] comment
section. Like you can have a laugh, but in some serious matters, maybe you
shouldn't believe that because all sorts of people make comments on it, so
you don't know who said that, or where it comes from.
15
See, Notley, T., Chambers, S., Park, S. and Dezuanni, M., Adult Media Literacy in Australia: Attitudes,
Experiences and Needs. Western Sydney University, Queensland University of Technology and University
of Canberra, 2021; Zhange, S. and Chan, E, ‘The way misinformation travels through diaspora communities
— including the Chinese diaspora — deserves more of our attention’, First Draft website, 11 December
2020; Sun, W., ‘How Australia’s Mandarin speakers get their news’, The Conversation, 22 November 2018.
acma | 13
While understanding these demographic differences is important to help governments
and industry design targeted interventions 16, it is also important to recognise that
virtually all Australians are susceptible to exposure and belief in misinformation. Based
on the research outlined above, such as the clear association between trust in doctors
and scientists and being ‘informed’ about COVID-19, improving access to authoritative
sources of news and information may be an appropriate remedy to misinformation.
Finding 1: Most Australians are concerned about, and have experienced, online
misinformation. Higher exposure is associated with heavy use of digital
platforms, disproportionately impacting younger Australians.
Finding 2: Access to authoritative and trusted sources of news and information is

an important mitigation against misinformation. Those that rely on social
media as a main source of news have a greater likelihood of being
misinformed about COVID-19.
2.2. Role of digital platforms

According to the N&MRC survey, over 91% of adult Australians use at least one digital
platform on a weekly basis. 17 The most popular platforms – Facebook, Google Search
and YouTube – are each used by more than half of the population every week. These
platforms are also where many Australians are consuming news and information about
COVID-19 (Table 1).
Table 1: Weekly digital platform usage, by general usage, and consumption of

news and information about COVID-19, Australia
# Platform General usage Consumption of COVID-19

news and information
1. Facebook 73% 46%
2. Google Search 54% 30%
3. YouTube 51% 22%
4. Messenger 42% 8%
5. Instagram 39% 16%
6. WhatsApp 23% 6%
7. Google News 17% 15%
8. Twitter 16% 11%
9. Snapchat 16% 4%
10. LinkedIn 13% 4%
11. Pinterest 11% 2%
12. TikTok 9% 3%
13. Reddit 9% 4%
14. Apple News 6% 5%
15. Bing 5% 3%
16. WeChat 3% 1%
16
A recent Australian study recommended that public health messaging around COVID-19 should seek to
target specific groups with higher misinformation beliefs, such as young people and those from culturally
and linguistically diverse communities, and to work with these groups to ensure appropriate tonality and
delivery of the message; Pickles, K. et al, ‘COVID-19 Misinformation Trends in Australia: Prospective
Longitudinal National Survey’, 23(1) Journal of Medical Internet Research, 2021.
17
14 | acma
Note: Consumption of COVID-19 news and information includes both active and incidental consumption.
News consumption by platform users

Those who used Apple News, Google News and Twitter were the most likely to have
consumed COVID-19 news and information in the last week while on the platform
(Figure 6). This is consistent with the general news-oriented nature of these services,
as opposed to the more social or community-oriented platforms with lower levels of
reported consumption like Snapchat, Messenger and Pinterest.
Figure 6: Consumption of news and information about COVID-19, among

platform users (%)
7 12
30 37
43 48 48 52 57 58 60
67 73 74
81 81
93 88
70 63
57 52 52 48 43 42 40
33 27 26
19 19
Total Consumption No consumption
Note: Consumption of COVID-19 news and information includes both active and incidental consumption.
Survey respondents were also asked whether their consumption was active (‘I used it
specifically to find news or information’) or incidental (‘I came across news and
information while on the platform for other reasons’). The ratio of active-to-incidental
consumption provides insight into what platforms users go on to find COVID-19 news
and information.
acma | 15
Figure 7: Ratio of active-to-incidental consumption among platform users who
had consumed news and information about COVID-19 (%)
Google Search 53 47
Bing 43 57
Apple News 41 59
Google News 40 60
YouTube 32 68
Twitter 31 69
Snapchat 29 71
Pinterest 29 71
WhatsApp 26 74
WeChat 25 75
Messenger 24 76
Reddit 23 77
LinkedIn 19 81
Instagram 18 82
Facebook 18 82
TikTok 17 83
Active consumption Incidental consumption
Figure 7 above shows that users of digital platforms were generally more likely to
incidentally come across news and information about COVID-19 than purposefully
seek it out on these services. The only exception to this was Google Search, with
slightly higher levels of active to incidental consumption (53% to 47%). This result is
unsurprising given that search engines respond to user-generated inquiries, noting
that Google is also how most Australians access broader online resources.
If I saw something interesting [on social media], I’d be like, ‘I’ll have a quick
read of that’. But I’m not actively seeking news, more kind of like inadvertently
coming across it on social media. Unless I'm really interested about
something, then I’ll use Google and maybe I’d read a few different websites to
get a consensus.
At the opposite end, services like TikTok, Facebook, Instagram and LinkedIn all had
very low ratios of active-to-incidental news consumption, at less than 20%. This
suggests a low awareness or reliance on these platforms as a source of news and
information about COVID-19.
16 | acma
As the most popular platform overall, and for the consumption of COVID-19 news and
information (Table 1), the low rate of active-to-incidental news consumption on
Facebook is particularly notable. Emerging research has found there to be a positive
link between the sharing of misinformation and the attention levels of users on social
media. 18 As users who passively or incidentally consume news and information are
less likely to focus on the accuracy of the content posted, platforms with lower ratios of
active-to-incidental news consumption may be at higher risk of facilitating the
propagation of misinformation.
I don't go to Facebook to look at information, I just go to socialise.

But the problem is that these media outlets exist on Facebook, and that's how
I end up getting information.
Misinformation by platform
Given the constantly shifting nature of misinformation, difficulties in assessing
falsehoods, and the challenges in accessing relevant data, it is not possible to quantify
the true scale and volume of misinformation in Australia. However, consumer surveys
and open-source network analysis can help provide insights into which platforms are
at higher risk and help provide a baseline for future assessments.
We asked Australians who reported seeing COVID-19 misinformation to identify on

which digital platforms they had seen it. Figure 8 below shows the split among all
survey respondents (green line), as well as the breakdown only among the users of
that platform (blue column).
Figure 8: Reported experience of COVID-19 misinformation by platform
50%
40%
30%
20%
10%
Among platform users Among all respondents
18
Pennycook, G., et al., ‘Shifting attention to accuracy can reduce misinformation online’, Nature, 592, 590-
595, 2021.
acma | 17
Facebook had the largest share of reported COVID-19 misinformation, with over a
third of all Australians who had reported seeing COVID-19 misinformation (35%)
identifying Facebook as the platform they saw it on.
This ranking holds true even when accounting for its much higher usage than any
other platforms. Among the 73% of Facebook users in Australia, just under 50% report
seeing misinformation. This suggests that Facebook has a disproportionately high
impact on the volume of online misinformation present in Australia.
WeChat has the second highest level of reported misinformation among its respective
users (47%). This is consistent with reporting on the widespread availability of
misinformation on the platform and is a notable finding given WeChat’s popularity in
certain CALD communities. 19 Unlike Facebook, however, weekly use of WeChat is
relatively low within the overall Australian community (3%). 20 As such, platforms that
are more widely used like Twitter, Instagram and YouTube are likely to have a greater
impact on the total volume of misinformation in Australia.
These survey results are broadly consistent with the findings from our network
analysis. Our researchers, We Are Social, compiled and analysed a sample of publicly
available Australian conversations posted on social media related to 4 misinformation
narratives (anti-5G, anti-vax, anti-lockdown and QAnon) over a 12-month period. This
was based on a list of popular keywords used by supporters of each misinformation
narrative (such as #5gmindcontrol, #covid19vaccineexposed, #masksdontwork and
#qaustralia), across Facebook, Twitter, Instagram, Reddit and YouTube. An overview
of these findings is at Table 2.
Table 2: Volume of mentions and interactions relating to key misinformation

narratives in Australia, by platform, April 2020 – March 2021
Potential
Platform Mentions Likes Comments Shares
impressions
Facebook 16,527 2,134,912 711,643 782,527 430,777,798
Twitter 45,149 23,459 36,152 80,789 121,694,338
Instagram 14,270 2,139,991 183,939 N/A 114,996,577
Reddit 461 17,467 7,791 N/A 30,193,932
YouTube 216 10,396 3,602 8,236 3,130,516
Source: We Are Social, Social media insights into how online misinformation and disinformation are being
spread across social platforms in Australia, May 2021 [unpublished]
19
See, for example, Xiao, B., T. Aualiitia, N. Salim and S. Yang, ‘Misinformation about COVID vaccines is
putting Australia's diverse communities at risk, experts say’ ABC News, 4 March 2021.
20
Noting the small number of respondents who used WeChat in the past week (n=81), we would also expect
a higher margin of error in these results.
18 | acma
This research highlights how a relatively small number of posts can generate high
levels of sharing and engagement, reaching significant numbers of people. Across the
5 major social media platforms examined, Twitter had the highest overall number of
conversations that mentioned misinformation keywords (Table 2). Facebook and
Instagram ranked second and third respectively, but posts on these platforms had
much higher levels of overall engagement and potential impressions than those on
Twitter, driven by accounts with larger audiences.
The We Are Social research does not constitute an exhaustive examination of

misinformation in Australia. Rather, the intent of this exercise was to provide high-level
trends and insights into selected misinformation narratives over time. The analysis is
heavily dependent on the list of chosen keywords for each narrative 21, and subject to
broader limitations around the availability and use of data. For example, some
platforms do not provide access to geographic information or include information on
content that has previously been removed. This may be one reason for YouTube’s low
number of mentions, despite appearing to play a significant role in conspiracy-driven
online communities. 22
Finally, social media platforms prevent researchers from examining any conversations
within private or closed groups or communities, irrespective of the size or reach of
these groups. While recognising the privacy rationale behind this decision, this creates
a significant gap in our understanding, and is an area of concern for many social
media researchers due in part to credibility amplification from false content being
shared by friends and family. 23
Alternative platforms for misinformation

With the strengthening of platform moderation activities by some platforms throughout
2020 and the first half of 2021 (see Appendix C), an important trend in the propagation
of misinformation has been the migration of some conspiracy-driven individuals and
communities from mainstream sites like Facebook and Twitter towards smaller, free-
speech oriented platforms with less rigorous content moderation policies.
In the We Are Social sample of 200 conspiracy-driven Facebook groups and pages,
there were 4,479 mentions of alternative social networks, with many of these posts
inviting users to join alternative social networks (Figure 9). 24 Among the range of
alternative platforms used by conspiracy-driven communities in Australia – including
Gab, Parler, Rumble and MeWe – the most popular platform by total number of
references is Telegram.
21
This approach may also capture a number of conversations that use identified misinformation keywords
but do not appear to be spreading misinformation or form part of a broader misinformation narrative –
including commentary, criticism and satire. We Are Social estimates this could represent up to 30% of the
sample; We Are Social, Social media insights into how online misinformation and disinformation are being
spread across social platforms in Australia, May 2021 [unpublished].
22
For example, YouTube represented 16% of the top 50 link sources among the sample of 200 conspiracy-
driven Facebook group and page accounts; We Are Social, Social media insights into how online
misinformation and disinformation are being spread across social platforms in Australia, May 2021
[unpublished].
23
See, for example, Tantuco, V., On Facebook’s messaging apps, false information spreads undetected,
unchecked, EU Disinfo Lab, 12 February 2021.
24
We Are Social, Social media insights into how online misinformation and disinformation are being spread
across social platforms in Australia, May 2021 [unpublished].
acma | 19
Figure 9: Examples of Facebook posts from conspiracy-driven communities,
inviting followers to join alternative platforms
In Australia, use of Telegram has grown rapidly. App store rankings for Telegram first
increased noticeably in early September 2020, likely related to the organisation of the
anti-lockdown ‘Freedom Day’ rallies. Australian downloads of the Telegram app then
peaked in early January 2021 for both Apple iPhone (reaching #6 in the iOS App
Store) and Android devices (reaching #3 in the Google Play Store). This was likely tied
to a number of concurrent events, including the US Capitol riots, the removal of Parler
from both app stores, and Donald Trump’s ban from Twitter and Facebook (Figure 10).
Figure 10: Telegram app store rankings in Australia, iOS and Android
1st
Twitter and Facebook
ban Donald Trump
App Store rank
Freedom Day
protests in
101st
201st
iOS (Apple) App Store Android (Google) Play Store
Source: SimilarWeb, App store rankings, 1 May 2020 to 30 April 2021.
Conspiracy-driven Telegram channels gain subscribers through the circulation of

channel lists, allowing for discoverability of channels from similar groups and like-
minded individuals. While Telegram conversations were not included in the We Are
Social quantitative analysis, these lists allowed researchers to identify a separate
sample of the most popular conspiracy-driven Telegram channels in Australia.
20 | acma
Former television chef Pete Evans appears to have the most popular conspiracy-
driven Telegram channel in Australia, posting an average of 12 posts per day to his
35,000 followers, with many of his posts containing unproven or objectively false
claims about COVID-19 cures and QAnon conspiracies.
While this pales in comparison to the number of followers Pete Evans had on
Facebook and Instagram before he was banned from these platforms in February
2021, his Telegram channel has more subscribers than almost all of the local
conspiracy-driven Facebook communities and pages identified by We Are Social. 25 As
discussed further below, this highlights the role of local influencers in amplifying
misinformation narratives.
Finding 3: Given its nature and the ongoing challenges in accessing relevant data,
the true scale and volume of misinformation in Australia is currently
unknown.
Finding 4: Australians report seeing the most amount of misinformation on large

platforms such as Facebook and Twitter. However, private messaging
services and smaller platforms with less strict content moderation
policies, like Telegram, are also being embraced by conspiracy-oriented
communities.
2.3. Sources and amplification

To assess overall attentiveness and understand more about the origins of online
misinformation, we asked Australians who had seen COVID-19 news and information
on social media whether they could recall or identify the source of the news or
information.
Social media posts from ‘official sources’ (such as the Australian Government or the
World Health Organisation) and ‘news media’ (such as the ABC or The Australian)
were the most popular sources of COVID-19 news and information overall. However,
those who were ‘highly misinformed’ about COVID-19 were more likely than the
‘informed’ respondents to get their information from less reputable sources, like
celebrities and social media influencers, links posted by people they know, or links
posted by strangers. 26
25
26
The Digital News Report: Australia 2021 also found important differences in news sources between
platforms, with social media personalities and influencers (21%) the second most common news source
among Instagram users, behind mainstream news outlets or journalists (23%); Park, S., Fisher, C., Lee, J.,
K. & McCallum, K., Digital News Report: Australia 2021, News & Media Research Centre, June 2021.
acma | 21
Figure 11: Sources of news and information about COVID-19 while on social
media (%)
68
Posts from official sources (e.g. WHO) 58
46
71
Posts from news media (e.g. ABC) 65
59
8
Posts from celebrities/influencers 14
17
29
Opinions from a person you know 29
35
11
Opinions from a person you don’t know 16
23
30
Links shared from a person you know 28
34
11
Links shared from a person you don’t know 15
24
2
I don’t notice 4
6
informed misinformed (low) misinformed (high)
Influencers and conspiratorial convergence

Conspiracy theories tend to start small and remain mostly insular, shared within
communities of people that already agree with one another, forming echo-chambers
that reinforce a particular worldview. 27 However, conspiratorial content can quickly
propagate to a wider audience via super-spreaders, who can draw audiences from
distinct communities into shared misinformation narratives. 28
In a recent large-scale internal study of vaccine-hesitant content on its platform,

Facebook’s data scientists identified half of this content originated from just 10 out of
638 population segments. Among the population segment with the most vaccine
hesitancy, only 111 users were responsible for half of all vaccine-hesitant content. 29
27
Sunstein, C. and Vermule, A., ‘Symposium on Conspiracy Theories: Causes and Cures’, Journal of
Political Philosophy, 17(2), 202-227, November 2009.
28
Starbird, K. et al, ‘Ecosystem or Echo-System? Exploring Content Sharing across Alternative Media
Domains’, Proceedings of the International AAAI Conference on Web and Social Media, 2018.
29
Dwoskin, E, ‘Massive Facebook study on users’ doubt in vaccines finds a small group appears to play a
big role in pushing the scepticism’, The Washington Post, 15 March 2021.
22 | acma
Throughout the pandemic, we have similarly seen a relatively small number of
Australian celebrities, sporting figures, politicians and prominent online influencers
exerting an outsized influence over COVID-19 misinformation narratives, while also
growing loyal, vocal and highly engaged groups of online supporters.
In our commissioned research, We Are Social compiled a list of the top 20 Australian
influencers sharing misinformation narratives, based on their total number of
interactions. Heading this list was celebrity chef Pete Evans, followed by Federal MP
Craig Kelly, and prominent anti-vaccine campaigner Taylor Winterstein. 30
COVID-19 conspiracy theories have proven to be particularly pervasive because they

have brought together people from different backgrounds and with very different
concerns. This includes members of the holistic health community, who are distrustful
of pharmaceutical companies and vaccines, anti-authoritarian groups who see
lockdowns as an attack on the freedoms of the individual, and followers of QAnon who
believe COVID-19 is a government cover-up.
There is considerable overlap between the 4 misinformation narratives examined by

We Are Social. From the list of the top 20 local influencers discussed above, 12 (60%)
had posted on topics relating to all 4 narratives. Similarly, more than half of the posts
in the We Are Social sample (54%) include misinformation keywords from at least 2
narratives. The largest overlap was between anti-lockdown and anti-vaccine
conversations (16.5%), followed by anti-lockdown and QAnon (8.8%). All 4 narratives
were referenced in 1.1% of posts (Figure 12).
Figure 12: Share of conversation by selected narrative within selected

conspiracy-driven groups and accounts, April 2020 to April 2021
Note: Based on share of conversation across a sample of 100 Facebook groups, 100 Facebook pages and
91 Instagram accounts. Diagram is illustrative and not proportionate. Does not equal 100% due to rounding.
30
acma | 23
Amplification techniques
Online misinformation is often characterised as a ‘rabbit hole’. The user journey may
start innocuously, such as by engaging with a piece of online content raising legitimate
concerns around vaccine safety. However, surrounding discussions often contain
pathways to misinformation, such as links to anti-vaccine conspiracy websites, or
invitations to join online communities of like-minded people.
There are a range of social and cognitive biases that drive people to engage with
misinformation, such as in-grouping, political fragmentation, and identity-based
conflict. Research suggests people may be driven to engage with misinformation on
ideological grounds or to defend their worldview, even if they do not believe the
content. 31
Content created for these communities is designed with virality in mind, using a variety
of techniques to promote engagement. One approach is to post material that fuels
outrage, such as unverifiable but highly intimate and emotive first-hand anecdotes. 32
Our research shows that the most common reaction among users engaging with
conspiracy-driven Facebook groups is ‘anger’ (Figure 13). By contrast, the most
common reaction to Facebook posts from mainstream Australian media outlets is
‘Haha’, and the most common reaction to posts from the top 745 most popular
Australian public figures is ‘Love’.
Figure 13: Share of Facebook reactions within conspiracy-driven groups,

Australian media and public figures, May 2020 to April 2021 (%)
2 4 3
8 9 3
3
8
21
19
54
21 Care
30 Wow
24 Sad
Love
35
33 Haha
21
2 Angry
Conspiracy-driven Australian media Australian public figures

groups
Note: ‘Conspiracy-driven groups’ consists of 100 Facebook groups from the We Are Social sample.
‘Australian media’ represents the Facebook pages of 49 mainstream media outlets in Australia. ‘Australian
public figures’ represents the 745 most-followed Australians on Facebook.
Another way to promote engagement and amplification of misinformation is by building

and strengthening a sense of community and shared belief. Members of these
communities see themselves as ‘free thinkers’ or ‘truth seekers’, who avoid
31
See, for example, Nefes. T. S., ‘The impacts of the Turkish government’s conspiratorial framing of the
Gezi Park protests’, Social Movement Studies, April 2017.
32
Glaser, A. and Zadrozny, B., ‘Distancing from the vaccinated: Viral anti-vaccine infertility misinfo reaches
new extremes’, NBC News, 14 May 2021.
24 | acma
mainstream news and rely on their communities for alternative news. 33 The use of
visual memes, symbolism and coded language in posts helps create a shared identity,
built around secret or hidden knowledge that the public is not privy to, or that people in
authority are trying to hide. 34
In this regard, some content moderation techniques adopted by digital platforms can
be ineffective or counterproductive in addressing misinformation. 35 Content removal or
de-platforming feeds into the general belief that platforms are involved in a deep-state
‘cover-up’. It also encourages members of conspiracy-driven communities to take
steps to pro-actively avoid detection or automated content moderation tools. We Are
Social found widespread use of intentionally misspelling keywords in posts, such as
‘v8ccine’ and ‘vackseen’. 36
Widespread content moderation by the platforms may also drive these conversations
further underground, by encouraging mass migrations to smaller alternative social
media or encrypted messaging apps. This makes it harder for governments,
researchers and platforms themselves to monitor and address harmful disinformation
and misinformation.
Role of news stories and the media itself

To further increase engagement and amplification, those seeking to spread
misinformation often reinterpret and leverage current events, political announcements
and other news of the day.
It is common for Australian conspiracy-driven Facebook communities to link to articles

and videos from mainstream Australian news sources (for example, Sky News, ABC)
and provide supporting commentary that is either critical of the news story or suggest
that it provides evidence or support of a broader misinformation narrative. Of the top
50 links shared on these group and page accounts, 40% were to mainstream news
websites, while another 23% were alternative news websites or blogs from unreliable
or untrusted sources (Figure 14).
33
34
Marwick, A., ‘Why do people share fake news? A sociotechnical model of media effects’, 2 Georgetown
Law Technology Review, 474 (2018). Researchers also note evidence of diversity and dissent in opinion
within conspiracy-driven communities, which is removed when content like memes shift across platforms,
decontextualising and amplifying false narratives; Krafft, P. M. and Donovan, J., ‘Disinformation by Design:
The Use of Evidence Collages and Platform Filtering in a Media Manipulation Campaign’, Political
Communication, 37:2, 194-214.
35
Smith, R. et al, Under the surface: Covid-19 vaccine narratives, misinformation and data deficits on social
media, First Draft, November 2020, p. 14.
36
acma | 25
Figure 14: Top 50 Facebook link sources from 200 Australian conspiracy-driven
Facebook group and page accounts, by source category
6%
4%
5% Mainstream news websites
6% Alt. news websites

40% YouTube
Alt. pages or groups (Facebook)
16% Politician pages (Facebook)

Alt. video sharing websites
Other
23%
In discussions with the ACMA, Australian news organisations have noted that they are
cognisant of this issue, particularly in regard to potentially misleading article headlines
and thumbnails used for video clips. Some outlets report they are now taking more
active steps to reduce the likelihood that their news content is misused or taken out of
context, like increasing social media training for journalists and other content creators.
Misinformation narratives can also be amplified by news media outlets themselves.

They can direct considerable attention to falsehoods and help these conversations find
a much wider audience, particularly where a story involves a celebrity or public
figure. 37 Focus group participants discussed the importance of the news media in
addressing or fact-checking misinformation but were also concerned about the impact
of the media in amplifying misinformation. Participants from separate groups also
made unprompted comments on the state of media diversity in Australia when asked
about who’s responsibility it was for combatting misinformation.
When I see misinformation or just blatant conspiracy theories being

perpetuated online and [then] actually make their way into the sort of
mainstream media discussion, and people are being exposed to them on a
wide basis, that makes me concerned. […] I mean, they should be addressed,
they definitely need to be addressed by the mainstream media and shut down,
but it's concerning that they're reaching millions and millions of people.
I think obviously it’s much, much easier for misinformation to be spread if

there's a limited diversity of new sources.
See, for example, Evanega, S. et al., ‘Coronavirus misinformation: quantifying sources and themes in the
37
COVID-19 ‘infodemic’, the Cornell Alliance for Science, 2020.
26 | acma
I think the biggest thing is making sure there’s media diversity so you get
information from all sorts of perspectives, instead of just being controlled in a
centralized way.
Male, 20s, Focus Group #4
International influence
Many misinformation narratives originate overseas but evolve or adapt to local
audiences and domestic issues. The best example is the QAnon conspiracy.
Originating as a series of cryptic messages on the imageboard 4chan in 2017, QAnon

was built on debunked far-right claims that Democratic political figures like Hillary
Clinton belonged to a cabal of Satan-worshipping paedophiles. Over time, this
conspiracy grew to include other ‘powerful elites’ like Bill Gates and Pope Francis, as
well as encompass a broader range of deep-state and anti-globalist rhetoric that could
be applied in local contexts. During Melbourne’s 2020 lockdown, for example,
Australian QAnon followers heavily promoted the theory that the state government
lockdown was a cover to traffic stolen children through secret tunnels under the city. 38
In 2020, Australia was considered the 4th most active QAnon country in the world,
responsible for approximately 2% of the global QAnon-related Twitter conversations. 39
QAnon was also the most popular of the 4 Australian misinformation narratives We are
Social examined over the 12-month period to March 2021, representing 31% of the
total conversations in the broader sample. 40 In the N&MRC consumer research,
several of the focus group participants shared QAnon-related anecdotes, with one
man talking about how his daughter had become a QAnon believer.
[…] when I hear her talking to her friends who I know, it’s like listening to, I
don’t know, Nazi propaganda. It’s just like where the frig do they find this stuff
out from? It’s all QAnon crap – we were talking about Trump a few months
ago – and she goes ‘oh he’s done some good things’ and I said, ‘what?’ and
she said, ‘oh against the paedophiles’ [group laughter] […] working out of a
pizza shop wherever it is in Washington somewhere, Hilary Clinton and eating
corpses now. It’s just nuts.
International supporters of misinformation narratives contribute to, and engage with,

local conspiracy-driven networks and communities. We Are Social estimate that
almost 60% of the identified misinformation conversations within their sample
contained global rather than local themes. Among Australian conspiracy-driven
communities on Facebook and Instagram, former US President Donald Trump was
mentioned 3 times more than Australian Prime Minister Scott Morrison, and 17% of the
Facebook pages in the We Are Social sample had international administrators, mostly
from the US, Canada and Israel. 41
Disinformation campaigns
In addition to promotion from genuine supporters, some conversations within
Australian misinformation communities appear to be spread inorganically, such as via
38
Kolankiewicz, V. ‘What lies beneath: tunnels for trafficking, or just a subterranean service? Time to rescue
these spaces from the conspiracists’ The Conversation, 14 September 2020.
39
Gallagher, A. et al., Key trends in QAnon activity since 2017, ISD, 2020.
40
41
ibid.
acma | 27
the use of bots or fake users. This suggests the presence of disinformation
campaigns, typically orchestrated by bad actors seeking financial gain, or by foreign
governments or other entities seeking to intentionally cause social harm by
undermining trust in a democratic process, the breakdown of community cohesion, or
the destabilisation of local institutions.
Facebook has publicly identified and removed 4 coordinated inauthentic behaviour

(CIB) networks directly relevant to Australia – 2 where Australia was a target of foreign
CIB networks, one where Australia was a country of origin in a CIB network targeting
foreigners, and one instance of domestic CIB. 42
Table 3: Facebook disclosure of known CIB networks of relevance to Australia
Date of public Domestic/ Overview

disclosure Foreign
8 March 2019 Domestic Facebook suspended the personal account of a

candidate running for NSW parliament after an aide set
up multiple fake accounts to smear a political rival. 43
26 March 2019 Both Australia was one of several countries targeted by a CIB
network, originating in Macedonia and Kosovo. This
network used fake accounts purporting to represent local
political communities and posted about religious and
political topics. 44
3 October 2019 Both Australia was one of several countries targeted by a CIB
network using fake accounts and localised content to
artificially increase engagement and promote content
related to the UAE. Facebook identified links to marketing
firms operating out of the UAE and Egypt. 45
6 August 2020 Both Facebook removed a CIB network that operated from
multiple regions, including Australia, linked to the digital
media outlet Truthmedia. 46
Source: Facebook, various CIB reports.
It is difficult to identify inauthentic coordinated activity, and even more difficult to

identify the sources behind it. Nevertheless, it is almost certain that Australians have
been the target of many more online disinformation campaigns than those published
by Facebook – particularly over the last 18 months.
In January 2020, for example, researchers at Queensland University of Technology

(QUT) discovered evidence of Twitter bots being used to amplify the narrative that
Australia’s 2019/20 bushfires had been deliberately lit. 47 In July 2020, researchers
from the Australian Strategic Policy Institute (ASPI) found reports of deaths from an
42
Facebook, Threat Report: The State of Influence Operations 2017-2020, May 2021.
43
For this matter, Facebook provided comments via local media reporting rather than publish details on its
website; Millington, B., ‘Port Stephens Liberal candidate Jaimie Abbott linked to trolling from fake Facebook
accounts’, ABC News, 9 March 2019.
44
Facebook, Removing Coordinated Inauthentic Behavior from Iran, Russia, Macedonia and Kosovo, 26
March 2019.
45
Facebook, Removing Coordinated Inauthentic Behavior in UAE, Nigeria, Indonesia and Egypt, 3 October
2019.
46
Facebook, July 2020 Coordinated Inauthentic Behavior Report, 6 August 2020.
47
Graham, T., and Keller, T., ‘Bushfires, bots and arson claims: Australia flung in the global disinformation
spotlight’, The Conversation 10 January 2020.
28 | acma
entirely fictitious US-led vaccine trial in Ukraine appearing on prominent Australian
anti-vaccination Facebook groups, days after Russia announced plans to develop its
own vaccine. 48 There have also been examples of disinformation used by state-
aligned actors to publicly criticise Australia, such as the well-publicised tweet from a
Chinese foreign ministry spokesperson in November 2020, publishing a doctored
image of an Australian solider holding a knife to the throat of a child. 49
Disinformation poses an ongoing threat to Australia. In addition to bots and troll farms,
a range of new technologies, underpinned by advancements in AI and machine
learning, continue to evolve and provide new tools for bad actors to intentionally
spread harmful falsehoods. 50
A number of initiatives are underway across the Australian Government to monitor and
respond appropriately to broader disinformation campaigns that can cause damage to
Australia’s interests or reputation (see Appendix E). However, it is also necessary to
recognise the range of downstream harms that can occur once this type of content is
widely shared by ordinary users. There is an ongoing role for government in
monitoring harmful disinformation and misinformation campaigns, as well as a broader
coordination role in bringing together stakeholders on how to best respond to this
content and protect Australians when online.
Finding 5: Misinformation typically stems from small online conspiratorial

communities, but can be amplified by influential individuals, digital
platform design, as well as the media.
Finding 6: Conspiratorial content is designed to be highly engaging, fuelling

outrage, and building on a sense of community. The confluence of
conspiracy theories around COVID-19 has created more paths to online
misinformation.
Finding 7: There is some evidence of co-ordinated inauthentic activity surrounding

popular misinformation narratives in Australia. Those who spread
misinformation often seek to reframe global conspiratorial narratives,
like QAnon, in a local context.
2.4. Impact and harms

Widespread belief in harmful misinformation can have serious impacts on individuals
and society, with the potential to cause a broad range of harms. These harms can be
acute, such as posing an immediate and serious threat to an individual’s health and
safety, or chronic, such as the gradual undermining of trust in public institutions or
authoritative sources of information.
Incitements to violence
While platforms have historically acted against content posing serious and imminent
threats of harms, the 2021 US Capitol riot is an example of the impact of longer-term
48
Thomas, E. et al., Pro-Russia vaccine politics drives new disinformation narratives, Australian Strategic
Policy Institute, 24 August 2020.
49
See, for example, Dziedzic, S. and Norman, J., ‘Scott Morrison demands apology from China over
'repugnant' tweet showing Australian soldier threatening to kill child’, ABC News, 1 December 2020.
50
Researchers, for example, have found AI generated algorithms are effective at automatically generating
short and believable messages on social media containing false narratives; Knight, W. AI Can Write
Disinformation Now—and Dupe Human Readers, Wired, 24 May 2021.
acma | 29
chronic harms arising from the widespread belief in misinformation, and how this can
spill over to the real-world as incitement to commit violent acts.
The fact that it has culminated in the storming of the Capitol, you know, it’s not
ridiculous – it’s a real thing.
On 6 January 2021, a mob of angry supporters of outgoing US President Donald

Trump, many with links to the QAnon conspiracy, stormed the US Capitol building in a
violent attack that left 5 dead and 140 capitol police officers injured. This insurrection
was not spontaneous. Rather, it had been widely discussed and organised on social
media, representing the culmination of the ‘#StopTheSteal’ misinformation narrative
that was first seeded online in September 2020 and escalated in the days and weeks
following the November election. 51
While platforms were quick to respond and quell calls for violence, Twitter’s CEO later
told a Congressional hearing that he takes some responsibility for allowing falsehoods
about the US election to spread on Twitter in the lead up to the 6 January rally. 52
Separately, an internal Facebook memo on the insurrection noted there were gaps in
Facebook’s policies around authentic coordinated activity, and noted that the
company’s focus on individual violations made them miss the larger harm that was
occurring across the network at this time. 53
Despite considerable efforts to disrupt and deplatform these networks, online

conspiracy theories still pose a real risk of violence. On 4 June 2021, the FBI issued a
bulletin warning that some self-identified QAnon adherents could seek to engage in
violence againgst political opponents, driven by a belief that they can no longer ‘trust
the plan’ and have an obligation to move from ‘digital soldiers’ to real-world action. 54
Impacts on individual and public health

There are also many clear and significant examples of the real-world consequences of
online misinformation locally, particularly in the context of the ongoing pandemic. A
current challenge for Australia concerns the propagation of anti-vaccine misinformation
narratives and growing vaccine hesitancy within the community. Based on the We Are
Social sample, Australian misinformation conversations peaked in March 2021, driven
almost entirely by growth in the anti-vaccine narrative over the previous 3 months.
A prominent 2014 study shows that exposure to anti-vaccine conspiracy theories can
directly affect vaccination intentions, introducing undue suspicion about vaccine
safety, while also decreasing trust in authorities. 55 Another more recent large-scale
global study of vaccine sentiment expressed on Facebook pages found that those who
51
DFRLab, #StopTheSteal: Timeline of Social Media and Extremist Activities Leading to 1/6 Insurrection,
Just Security, February 2021.
52
Conger, K., Jack Dorsey says Twitter played a role in U.S. Capitol riot, The New York Times, 25 March
2021.
53
Mac, R. et al, Facebook Stopped Employees From Reading An Internal Report About Its Role In The
Insurrection. You Can Read It Here, BuzzFeed News, 26 April 2021.
54
Benner, K. The F.B.I. warns that some QAnon believers could turn to violence as predictions fail to bear
fruit, The New York Times, 15 June 2021.
55
Jolley, D. and Karen, D., The Effects of Anti-Vaccine Conspiracy Theories on Vaccination Intentions,
PLOS One, 9, e89177, 2014.
30 | acma
were undecided on vaccines tended to interact more heavily within clusters of pages
sharing anti-vaccination views, compared to those with pro-vaccination views. 56
In the context of both ‘data deficits’ and the over-supply of information surrounding
vaccine technology, trials and side effects, bad actors can take advantage of the
resulting uncertainty to spread anti-vaccination misinformation to more Australians and
increase levels of vaccine hesitancy. 57 Concerns about vaccine misinformation were
shared by most – but not all – of our focus group participants.
But what worries me is that you might get someone old that sees that.
That thinks this vaccine is not safe for me. It might put them off getting
vaccinated. Because it's just these false claims that they put out and
vulnerable people could just believe it, you know?
[...] having this misinformation around the internet and so accessible

is actually creating fear and danger in communities because it means that
people won’t get vaccinated. They don't believe what politicians, what
scientists are saying, like, ‘this is what we're going to do to get over this
pandemic and get through it’.
Female, 18, focus group #12
It's well within our rights to be questioning it every step of the way, and we
should be. It doesn't mean we're conspiracy theorists […] I totally 100%
understand and believe that COVID is a real thing […] But sometimes I just go
‘what if it’s more the other way, and that the government, not that they're
faking it, but that they're exercising a kind of overt control that they maybe
don’t need to be doing?’. […] And suddenly, we're all having jabs and shit.
Financial impacts
While it is difficult to quantify, misinformation also has a financial impact. A 2019
economic study estimated the annual global cost of fake news at US$78 billion, driven
by a range of factors including market volatility and losses, financial scams, public
health costs, expenditure on political races, and reputational damage to brands. 58 This
study, which pre-dates COVID-19, helps highlight the pervasive nature of
misinformation and its growing impact across all sectors of the economy.
Locally, mobile carriers in Australia have recently incurred a range of new and
unforeseen costs related to the rise and propagation of 5G and electromagnetic
energy (EME) misinformation. These costs, estimated at A$7.9 million in 2020, are
discussed further in the Impacts of anti-5G misinformation case study below.
56
Johnson, N. et al, The online competition between pro- and anti-vaccination views, Nature, 582, 230–233,
May 2020.
57
A data deficit is where there are high demands for information on a topic, but credible information is in low
supply – either because it doesn’t exist or isn’t reaching its intended audience. In a study of social media
conversations, First Draft identified a range of key data deficits relating to vaccines, resulting in increasing
vaccine scepticism; Smith, R. et al, Under the surface: Covid-19 vaccine narratives, misinformation and data
deficits on social media, First Draft, November 2020.
58
Cheq, The Economic Cost of Bad Actors on the Internet – Fake News, 2019.
acma | 31
Case study: Impacts of anti-5G misinformation
While the rollout of new wireless networks is occasionally controversial, COVID-19 has
served to escalate worldwide opposition to 5G technology and deepen community
concern about the negative health impacts of electromagnetic energy (EME).
From late January 2020 onwards, a growing number of online conversations appeared
linking the emergence of COVID-19 to pre-existing conspiracy theories about 5G.
These ranged from allegations that 5G weakened the immunity system, to claims the
virus was either spread directly by 5G, was fake and covering up 5G symptoms, or
had been intentionally released to hasten the 5G rollout. Some of these conversations
appeared organically, while others showed evidence of bot-like behaviour or other
techniques common to coordinated disinformation campaigns. 59
These conspiracies rapidly gained traction during the initial stages of lockdown, further
amplified by posts from celebrities like US actor Woody Harrelson, UK boxer Amir
Khan and US rapper Wiz Khalifa. 60 By early April, public concerns had escalated
across Europe and the UK, resulting in reports of telecommunications engineers being
threatened or harassed, and arson or vandalism attacks on at least 30 mobile towers
in Britain. 61 During this time, an Essential Poll showed 12% of Australians believed
that the 5G network was being used to spread the COVID-19 virus 62, and membership
of Australia's largest anti-5G Facebook group surged from 6,800 to more than 48,000,
fuelled by local celebrities and influencers appealing to local concerns. 63
Of the 4 key misinformation narratives monitored to inform this report, anti-5G had the
lowest overall share of mentions over the last 12 months. As per Figure 15 below,
posts with anti-5G keywords reached a peaked in mid-2020 in response to small anti-
5G rallies being held across Australia, but quickly dissipated relative to the other
narratives. This sharp decline coincided with a range of public communications
campaigns across government and industry, as well as actions taken by Facebook
and other platforms to limit the reach of content linking 5G to COVID-19.
Figure 15: Volume of mentions by key misinformation narratives
3k
Anti-vax and 5G
2k protest rally in Sydney
Mentions
and Melbourne
1k
Anti-5G Anti-vaccine Anti-lockdown QAnon
59
Gallagher, R. 5G Virus Conspiracy Theory Fueled by Coordinated Effort, Bloomberg News, 9 April 2020.
60
Bruns, A., et al., ‘Corona? 5G? or both?’: the dynamics of COVID-19/5G conspiracy theories on
Facebook, 177(1), Media International Australia, 12-29, November 2020.
61
Satariano, A. and Alba, D., Burning Cell Towers, Out of Baseless Fear They Spread the Virus, The News
York Times, 10 April 2020.
62
Essential Research, Belief in Conspiracy Theories, 19 May 2020.
63
Nicholls, S., et al. What is the truth about 5G? Four Corners spoke to leading experts and anti-5G activists
to find out, ABC News, 3 August 2020.
32 | acma
While there are still highly localised pockets of anti-5G sentiment – such as in Northern
NSW – the research suggests that Australia’s efforts to combat 5G misinformation
have been mostly successful. 64 Yet despite its relatively short-lived nature, there is a
considerable financial cost associated with the spread of 5G misinformation narratives,
largely borne by governments and industry, and ultimately passed on to consumers.
In late 2020, the ACMA approached the Australian Mobile Telecommunications

Association (AMTA) to gauge whether the mobile industry would be interested to
participate in an exercise to quantify the cost of addressing 5G and EME
misinformation. Telstra, Optus, TPG Telecom and AMTA all provided the ACMA with
high-level cost inputs, allowing us to estimate the financial impact across the industry.
Based on the industry data collected by the ACMA, the Australian mobile industry
spent an estimated A$11.0 million across calendar years 2019 and 2020 as a direct
result of misinformation about EME and/or 5G. Most of these costs were incurred in
2020, with total annual industry expenditure more than doubling from 2019 – from
A$3.1 million to A$7.9 million.
Figure 16: Estimated cost to the Australian mobile industry of addressing EME
and 5G misinformation
$4M +550%
+274%
$2M -21%
+7% +467%
$M
Repairing or Additional rollout Public education Complaints Other costs
replacing damaged costs campaigns handling
property procedures
2019 2020
Source: ACMA, based on cost estimates provided by Telstra, Optus, TPG Telecom and AMTA.
All 3 Australian carriers reported instances of arson or vandalism attacks at mobile

sites that were related to the propagation of 5G/EME misinformation. While the total
number of impacted sites in Australia was much lower than in other countries,
replacement or repair costs at these sites nevertheless represented the largest cost
category in this exercise. Carriers estimate they spent A$3.3 million in relevant repair
costs during 2020, representing a 550% increase on 2019. While not captured in the
cost data, this remains an ongoing issue for some carriers, with site vandalism due to
5G/EME provocation continuing throughout 2021.
Carriers also reported a 274% increase in additional rollout costs, due to new functions
like staff training to counter aggressive behaviours, and security measures at sites.
Other costs in 2020 were driven by monitoring and research on misinformation, and
the decommissioning and relocation of sites due to community backlash.
There are a range of larger, intangible costs associated with 5G misinformation not
included in these calculations. These include the cost from delays to the rollout of 5G
networks, and related reduction in productivity benefits across society. An AMTA-
commissioned study from 2019 projected the mobile sector, underpinned by 5G, to be
worth A$65 billion to the Australian economy by 2023. 65 AMTA noted a portion of
these productivity benefits are unlikely to be realised due to 5G misinformation.
64
Particularly when compared to other countries, such as India; Here’s why ‘5G spreads Covid’ is a myth,
says government, The Times of India, 5 June 2021.
65
Deloitte Access Economics, Mobile Nation 2019: The 5G Future, 2019.
acma | 33
Finding 8: Misinformation narratives can result in a wide range of acute and
chronic harms, including the erosion of trust in authoritative sources and
democratic institutions over time.
Finding 9: The real-world consequences of misinformation have been readily

apparent over the past 18 months: inciting violence, undermining official
health advice, and causing tangible financial impacts on governments,
industry and consumers.
2.5. Platform measures and accountability

Most digital platforms have a range of existing measures to address seriously harmful,
false or misleading information, and many of these policies, tools and initiatives have
been strengthened over the previous 18 months in the context of COVID-19 (see
Appendix C).
Ahead of the commencement of the code, N&MRC asked Australians about their
knowledge of these existing efforts. There was a general awareness of some of the
measures, but very few Australians reported having directly seen or experienced any
of these measures (Figure 17).
Figure 17: Awareness and experience of platform measures
Removal of misinformation 7 48 44
Labelling of misinformation 13 39 48
Providing options to report misinformation 10 42 48
Directing users to authoritative sources 11 34 55
Making trusted information more visible 9 32 59
Directing users to resources on the platform 9 30 61
I have seen or experienced it

I am aware, but have not seen or experienced it
I am not aware of it
Views on platform measures

The most recognised measure – removal of content – was also the most controversial.
Focus group respondents were split on the question of where platforms should draw
the line. Some were of the view that all legal speech should be allowed, and only be
hidden or removed if it was ‘pushing illegal activity’. Others disagreed, with one
participant arguing that ‘if it’s fake, it should be off the internet’. The most common
position was that there are times where it may be appropriate to remove content, but
platforms should not take this step lightly given these actions place limits on speech.
34 | acma
These discussions were framed by media reporting on events at the time of the
research, including the decision of Twitter and Facebook to ban the US President
Donald Trump in January 2021 following the US Capitol riots, and Facebook’s decision
to temporarily ban Australian news in February 2021, immediately before the passage
of the News Media and Digital Platforms Mandatory Bargaining Code (the news media
bargaining code). Participants expressed some concern about the amount of power
platforms wield in making these decisions, and the potential unequal application of
their policies.
I think it's a really good thing that Twitter censored [US President Donald
Trump]. But then, in saying that, it's also a really interesting thing of now that
they're censoring news in Australia, we’re like, ‘whoa whoa, you don't have
that right’. It's an interesting argument to have because when it's not
benefiting me, when I want to see my news on my news feed, then I'm like,
well no.
Yeah, I think in that case, it was for the safety of people and the safety of their
country. So I sort of get why they did it, but I could understand why people
would think no, you shouldn't be silenced. Who makes that decision?
There are key demographic differences in awareness of platform measures, which

appear closely aligned to the characteristics of heavy users of digital platforms – those
who were younger, male, and with high education were all more likely to be aware of
platform measures compared to the general population. 66 Interestingly, there were
also key differences between the users of particular platforms. Twitter users, for
example, were considerably more likely to be aware of platform measures than
Facebook users (Figure 18). This could reflect the greater publicity around Twitter’s
efforts, better messaging by Twitter to its users, or demographic differences in its user
base.
Figure 18: Awareness and experience of platform measures, by users of

platform (%)
75 75 72
64 62 62 65
58 55 56 55 57 58
48 50 48
44 42
Removal of Labelling of Providing Directing users Making trusted Directing users

misinformation misinformation options to report to authoritative information more to resources on
misinformation sources visible the platform
Facebook YouTube Twitter
66
acma | 35
Effectiveness of platform measures
While this chapter does not seek to assess whether specific, existing measures have
been effective, some high-level findings may be drawn from the research.
Although there was no noticeable decline in overall volume of misinformation

conversations in the We Are Social sample over the 12-month period, conversations
about certain misinformation narratives – like QAnon and anti-5G – did decline
dramatically over time, in line with stronger moderation activities on these topics.
Further, by examining posts from the sample of conspiracy-driven Facebook groups

over 12 months, we can see evidence of reduced amplification. The total volume of
posts within these groups fell by 91% over the year ending 31 March 2021, after
reaching a peak in September 2020 (Figure 19). Over the same period, there was a
192% reduction in the number of comments, 244% reduction in the number of shares,
and a 302% reduction of link interactions. 67
Figure 19: Number of posts by selected conspiracy-driven Facebook groups;

1 April 2020 to 31 March 2021
60K
40K
Post count
20K
Collectively, these findings reflect a limited snapshot, and more needs to be done to
better understand what measures work and how to monitor the effectiveness of
platforms’ moderation activities in Australia. Many researchers remain concerned that
platforms are inconsistent in their approaches to detecting and addressing harmful
misinformation and lack formal structures for information sharing and collaboration.
Misinformation is a shared responsibility

When asked about who bears the responsibly for addressing online misinformation,
most Australians see this as an issue of shared responsibility, with 78% agreeing that
‘individuals should use common sense and learn to detect false or misleading
information themselves’, and 76% agreeing that ‘platforms should be doing more to
reduce the amount of false or misleading information people see’ on their services
(Figure 20).
67
36 | acma
Figure 20: Responsibility to address misinformation (%)
Individuals should use common sense and learn to

78 15 4
detect false or misleading information themselves
Social media or online platforms should be doing

more to reduce the amount of false or misleading 76 17 5
information people see on these services
It is the government’s responsibility to make sure the

public is not exposed to false or misleading 59 25 12
information on social media or online platforms
False or misleading information is unavoidable and it

44 25 29
is just something we must live with
It is not the job of social media or online platforms to

decide what is or is not false or misleading 26 24 46
information
Agree Neither Disagree Don't know
A smaller majority (59%) agreed that ‘it is the government’s responsibility to make sure
the public is not exposed to false or misleading information on digital platforms’. This
result may reflect a general sentiment, communicated by several participants in the
focus groups, that platforms do not have sufficient commercial incentive to move
voluntarily on this issue.
I think, again, in a perfect world people just wouldn't use Facebook for news
but in the world we live in, I think […] the government has to have some input
into what is promoted to such a large group of people, especially because as
[participant] said, Facebook doesn't exist for the people. It exists for the
shareholders, it exists to generate profit, and increase its value. And I just I
can't see any world where leaving it alone ends with the betterment of media
and information.
There was considerably less consensus on whether platforms should be the ones to
decide on what is misinformation. Only a quarter of Australians (26%) agreed with the
statement that it was not the job of social media or online platforms to decide what is
or is not false or misleading information, with 46% disagreeing and saying it was.
Broken down, this result differed starkly between those who were ‘informed’ about
COVID-19 (15% agree) versus those who were ‘misinformed (high)’ (69% agree).
Further, while many see a role for government in addressing misinformation, focus
groups were also concerned that government could be perceived by some to be
untrustworthy, exercising excessive power, or impeding on individual freedoms of
speech. This is an important consideration for any future regulatory reforms.
acma | 37
I think if it’s something that has been debunked, or that is affecting a large
group of the population, it will probably be the government's job to at least put
that information out there. […] I really wish that there were NGOs, and I'd love
to support them, if they wanted to take this up because I don't like the
government having so much power either.
Finding 10: Most Australians are aware of platform measures to remove or label
offending content, but few have direct experience. Early evidence
suggests these steps have been somewhat effective in reducing
amplification of misinformation on particular platforms.
Finding 11: Australians see the issue of misinformation to be one of joint

responsibility – split between individual users, platforms, and
government. There is some scepticism in the ability of platforms to self-
regulate, and concern about government’s role in regulating speech.
Finding 12 Information on the effectiveness of platform measures is limited, and

more needs to be done to better understand what measures work and
to monitor the effectiveness of platform moderation activities.
38 | acma
3. Code development
This chapter provides an overview of the code development process and an
assessment of whether it has met the expectations set out by the government.
Findings in this chapter are based on the ACMA’s observations and expertise in code
development, as well as ongoing discussions with DIGI, code signatories, and other
key stakeholders we have engaged with over this period.
Table 4: Timeline of key events in code development
Date Event
December 2019 Government releases its response to the ACCC’s Digital

Platforms Inquiry, calling on digital platforms to develop a code
(or codes) on disinformation and news quality
January – March 2020 ACMA writes to, and meets with, most major digital platforms
in Australia, seeking information on existing measures and to
discuss expectations for code development
March 2020 DIGI informs the ACMA that it will take a leadership role in
developing an industry-wide disinformation code and will seek
involvement from both members and non-members
May 2020 DIGI provides the ACMA with a project plan. The plan signals
adoption of the code by November 2020
June 2020 ACMA releases its position paper, providing advice to industry
on code development, proposing a code model, and
formalising its expectations for the code
October 2020 DIGI releases a draft code for public consultation. Consultation
is open for 5 weeks. DIGI hosts a targeted stakeholder
roundtable
February 2021 DIGI hosts a second stakeholder roundtable for submitters.

DIGI publishes the final code, stakeholder submissions and a
summary report, and announces the initial 6 signatories
acma | 39
3.1. Project timeframes
The Australian Communications and Media Authority (ACMA) will oversee the
development of the code (or codes) and will report to the Government on the
adequacy of the platforms’ measures and the broader impacts of
disinformation with the first such report due no later than June 2021.
Government Response and Implementation Roadmap

for the Digital Platforms Inquiry, December 2019
Position 6: The ACMA expects …the code to be in place by no later than

December 2020.
ACMA, Misinformation and news quality on digital platforms in Australia:

A position paper to guide code development, June 2020
In its December 2019 Implementation Roadmap, the Australian Government set an

expectation that there would be a code (or codes) in force by the end of 2020. This
would provide the ACMA with at least 6 months to both review the code’s initial
operation and develop a report to government on the adequacy of platforms’
measures.
In the first quarter of 2020, the ACMA held discussions with the industry representative
body DIGI as well as a range of individual digital platforms (both members and non-
members of DIGI). These discussions focused on platforms’ existing disinformation
policies and measures, the process of developing a code, and what elements could be
included in a code.
In March 2020, DIGI indicated that it would lead drafting of an industry-wide code and
had engaged the University of Technology Sydney's Centre for Media Transition
(CMT) and First Draft to assist in this project.
In June 2020, the ACMA released its paper Misinformation and news quality on digital
platforms in Australia: A position paper to guide code development. This was designed
to assist industry in developing a code by articulating our expectations for both the
code contents and code development process, including on timelines and
consultation. The paper reiterated the government’s expectation that a code be in
place by the end of 2020. It further proposed signatories publish individual annual
action plans in January 2021 stipulating how they will meet their obligations under the
code and provide a progress report to the ACMA in April 2021 to inform our report.
DIGI released a draft code for public consultation in October 2020. The final code was
released publicly on 22 February 2021. At this time, DIGI announced 6 initial
signatories and noted it would formally be taking on the role of code administrator.
ACMA commentary
We understand that code development can be a lengthy and difficult process,
particularly for the first time. Given the complexity of the topic and the diversity of
potential signatories, the work undertaken by DIGI and industry members on drafting
an industry-wide code warrants recognition.
DIGI took on the responsibility of managing the project and successfully delivered on a
final code that was adopted by a wide range of both DIGI member and non-member
platforms. This work was undertaken in a complex environment with competing views
from within and across industry, government, and civil society, and had to be
completed to a relatively compressed timeframe for a first-time code.
40 | acma
Despite the overall achievement, it should be noted that the delay in finalisation of the
code was unfortunate, albeit understandable.
It is evident that DIGI, industry members and its CMT partners understood the
expected timeframe for the commencement of the code, but that development took
longer than initially anticipated. DIGI has explained that its original project timeline had
not accounted for the release of the ACMA position paper or the expectation that it
should consult with a diverse range of potential signatories before the draft code’s
release. DIGI also attributed the delay to the difficulties negotiating with several non-
DIGI members who had expressed interest in participating but needed more time to
review.
The ACMA appreciates this was a novel and complex project and it takes time to
reach consensus across an industry. We also acknowledge there were a range of
competing priorities throughout 2020. These included platforms’ evolving responses to
the COVID-19 pandemic, and several concurrent streams of work arising from the DPI,
including the introduction of the news media bargaining code.
However, it should be noted that this short delay significantly reduced the ACMA’s
ability to monitor and report on the effectiveness of the final code and prevented
consultative work with the industry on the ACMA’s proposed monitoring framework.
This delay, along with the deferral of code administration matters discussed below in
Chapter 4, should be recognised as a general limitation of this assessment process.
Finding 13: In leading code development, DIGI successfully managed a novel,

complex and time-sensitive project, navigating a range of competing
interests across a disparate group of stakeholders that included both
members and non-members of DIGI.
3.2. Public consultation process

Position 6: The ACMA expects platforms to undertake an open, public
consultation process when developing the code…

Transparency can encourage genuine dialogue and build trust in the policy
process, but in order for your consultation to be credible and effective, you
need to engage with stakeholders in a way that is relevant and convenient for
them. You also need to give stakeholders time to consider the information you
give them and time to respond.
The Australian Government Guide to Regulation, March 2014
In the June 2020 position paper, the ACMA set out its expectation that digital platforms
undertake meaningful public consultation on the drafting of the industry code. We
noted that input should be sought from experts across academia, relevant government
agencies, and impacted stakeholders including consumer groups and users of digital
platforms.
DIGI undertook a 5-week public consultation process on their draft code, supported by
a discussion paper, which was authored by CMT and provided additional background
research. DIGI invited submissions from 42 organisations they had identified as
relevant and ran a virtual roundtable discussion with 12 targeted academics and
relevant subject matter experts.
acma | 41
The consultation process generated 17 public submissions from a range of
stakeholders including academics, thinktanks, news media, and professional and civil
society organisations. These submitters were later invited to a briefing by DIGI on how
their feedback impacted the final code ahead of its release.
DIGI published these submissions on its website on 22 February 2021, coinciding with
the release of the final code. DIGI also prepared and published an 18-page summary
report that discussed how stakeholder feedback was addressed in the final code.
Submitter views on the consultation process

A number of submitters spoke favourably of their experiences
engaging with DIGI in this process, noting they ‘felt heard’ by DIGI
and welcomed the opportunity to discuss their views in a constructive
roundtable format with subject matter experts. They observed that
DIGI appeared to take the process seriously and had been
responsive to stakeholder concerns, evidenced by the significant
changes made between the draft and final code.
Other submitters were more critical of DIGI and questioned whether

the consultation had, in fact, been genuine. They raised concerns
about the limited participation in the first roundtable, and the lack of
consultation on the final code, noting it was too late to share their
views during the second roundtable as it was clear that nothing
would change as a result. Others noted difficulties contacting or
engaging with DIGI and observed that the summary report had failed
to fully respond to the issues raised in their submissions.
Some feedback on the consultation process was not directly relevant

and therefore not considered by the ACMA. For example, several
stakeholders advocated for a mandatory code and argued that DIGI
did not sufficiently respond to these calls.
ACMA commentary
DIGI ran a robust and meaningful consultation process.
It proactively identified and reached out directly to a range of relevant organisations to

inform them about the consultation and provided a reasonable timeframe to accept
submissions. The process attracted comments from a representative cross-section of
interested stakeholders, and DIGI provided a high level of transparency over how this
feedback fed into the final code.
However, there were certain aspects of the consultation that could have been
improved.
Publicity of the consultation process

The ACMA expected DIGI to undertake a full and open consultation on the code. On
release of the draft code, despite reaching out directly to a wide range of stakeholders,
DIGI made little effort to attract a broader range of public views. It did not put out a
media statement, post an update on its social media channels, or pursue any other
public communication avenues to increase awareness of the consultation process
beyond those organisations it had originally identified.
Given the breadth of concern regarding COVID-related misinformation in 2020, it was

important that consultation on this issue be as open as possible, particularly during the
code development period.
42 | acma
Noting that a key focus of the code is to provide more transparency, the ACMA
considers more should have been done to canvass a wider range of views across the
community. This includes among relevant health experts, organisations representing
CALD communities, a broader cross-section of media organisations, and users of
major platforms.
Limited opportunity for further engagement

Secondly, the ACMA is concerned there was a lack of opportunity for stakeholders to
provide comment on the content and drafting of the final code. A recurring observation
among submitters was that that they would have preferred if DIGI had engaged earlier
or sought additional comments on a final draft before launching the code.
This is of particular importance given the expanded scope of the final code compared
to the draft code. While signatories can rightly highlight this as evidence of being
responsive to stakeholder feedback, the inclusion of ‘misinformation’ was a significant
change. One submitter told the ACMA that it knew of other organisations that would
have made a submission had DIGI consulted on a code that included misinformation.
Another view expressed by some submitters was that the written submission process
was insufficient, and DIGI could have explored other avenues to seek feedback from a
more diverse group of stakeholders, such as via public meetings.
Some of these concerns could have been mitigated if DIGI had consulted earlier and
provided further opportunity to review and comment on the final draft before its
finalisation. However, we recognise that DIGI was facing timing constraints and that
this would have further delayed the release of the code. In these circumstances, DIGI
could have instead considered holding its second roundtable session earlier, along
with the earlier release of the public submissions, so that stakeholders could provide
additional verbal feedback ahead of the code’s finalisation.
Finding 14: DIGI undertook a meaningful public consultation process on its draft
code, generating a range of feedback from academia, industry, and
parts of civil society, which visibly informed the final code.
Finding 15: DIGI could have improved its consultation process with greater publicity,
including promoting it through existing public communications channels
and engagement with the media.
Finding 16: DIGI dealt with stakeholder feedback in a relatively open and
transparent manner. However, the significant change in scope meant it
would have been best practice to provide stakeholders a further
opportunity to comment on the final drafting prior to finalisation.
3.3. Code signatories

Position 3: The ACMA expects that the code will cover online search engines,
social media platforms and other digital content aggregation services with a
major presence in Australia. The ACMA would encourage all platforms,
regardless of size, to consider signing up to the code.
In line with the government’s DPI response, the ACMA indicated that the code should,
at a minimum, cover all ‘major digital platforms’ (defined by the ACCC to be search
engines, social media platforms or digital content aggregators with at least 1 million
active monthly users).
acma | 43
Given the voluntary nature of the code, the ACMA also encouraged all platforms,
regardless of size, to consider signing up to the code. We noted that steps to address
online misinformation could also be relevant to a range of other online services that
distribute news and information to Australians, including smart devices, online forums,
podcast aggregators and closed group messaging services.
Facebook, Twitter, Google, Microsoft, TikTok and Redbubble were named as initial
code signatories on release of the final code on 22 February 2021. In May 2021, DIGI
announced that Adobe and Apple had also signed up to the code.
Table 5: Code signatories

DIGI EU code
Signatory Covered service(s) 68 Start
member signatory
Facebook
Facebook
Instagram
22 Feb 2021  
Google Search
Google Google Ads 22 Feb 2021  
YouTube
Company-wide initiatives for

Microsoft its consumer-facing products 22 Feb 2021 
(including Bing and LinkedIn)
Redbubble Redbubble 22 Feb 2021 
TikTok TikTok 22 Feb 2021 
Twitter Twitter 22 Feb 2021  
Apple Apple News 25 Feb 2021
Adobe Content Authenticity Initiative 20 Apr 2021
The code does not apply to all products and services of signatories. As discussed in
Chapter 4, it is limited to those that deliver, to end users in Australia, user-generated
content (including sponsored and shared content) and/or content that is returned and
ranked by Search Engines in response to user queries. The code does not ordinarily
apply to news content, private messaging services, email, or enterprise services.
ACMA commentary
Based on publicly available reporting on the number of active monthly users, and our
assessment of web traffic, app store rankings and survey data, it is evident that the
code covers almost all major digital platforms in Australia.
We particularly welcome the inclusion of platforms that are not existing DIGI members
(Microsoft, TikTok, Apple and Adobe), and platforms that are not signatories to the
European Union (EU) Code of Practice on Disinformation (Apple, Adobe, and
Redbubble). This reflects the efforts of DIGI in seeking broad code coverage, and
68Note these services are drawn from signatory opt-in nomination forms and initial annual reports,
published online by DIGI on 22 May 2021.
44 | acma
signals strong commitment across the entire industry to voluntarily address the issue
of online disinformation and misinformation in Australia. All code signatories should be
commended for their decision to participate in this scheme.
However, the ACMA remains concerned about the lack of transparency surrounding
the signing of new digital platforms after the code’s commencement date. Apple, in
particular, signed up to the code on 25 February 2021, 3 days after the code came into
force. Disappointingly, Apple did not publicly acknowledge its involvement until DIGI
released the initial signatory reports 3 months later, on 22 May 2021. DIGI should
ensure that any future signatories to the code are prepared to announce their
participation as soon as practicable after signing.
Non-signatories
Despite the broad coverage of the code, there are a small number of popular social
media or search engine services that have not participated in this process. This may
be due to a lack of interest or perceived relevance, limited regulatory capacity or
expertise in the region, or a lack of awareness of the process. Table 6 provides a list of
the most popular non-signatory services by a variety of usage metrics.
Table 6: Usage metrics of key services not covered by the code

Digital Usage in Usage in Unique Total Top
platform past past week 70 monthly monthly Android
service month 69 website website apps by
visitors visits (excl. usage
(excl. apps) 72 rank 73
apps) 71 (million)
(million)
Snapchat 29% 16% 0.5 0.6 4
Pinterest 26% 11% 6.1 9.5 85
Reddit 19% 9% 6.9 58.6 44
Twitch 13% - 1.1 17.0 156
Tumblr 10% - 1.4 8.2 -
Discord - - 1.1 8.5 39
DuckDuckGo - - 0.9 23.8 166
Note: Data on unique monthly website visitors is not comparable to monthly active users and does not
capture usage of mobile apps. Survey data not available for all platforms.
69
Based on the GlobalWebIndex Q3 2020 survey of approximately 16,000 Australian internet users aged
between 16-64; We Are Social, Digital 2021: Australia, January 2021.
70
Based on a survey of 2,659 adult Australians between December 2020 and January 2021; N&MRC,
COVID-19: Australian News & Misinformation Longitudinal Study, 2021 [unpublished].
71
Sum of all unique website visits, from Australia, during March 2021. Includes domain and all meaningful
subdomains, across both desktop and mobile web; SimilarWeb, Top Websites – Custom Industry/Unique
Visitors/Australia.
72
Sum of all non-unique website visits, from Australia, during March 2021. Includes domain and all
meaningful subdomains, across both desktop and mobile web; SimilarWeb, Top Websites – Custom
Industry/Monthly Visits/Australia.
73
‘Usage rank’ is calculated by a SimilarWeb algorithm that factors in ‘Current Installs’ and ‘Active Users’,
providing a ranking of the top 1,000 free Android apps for the last 28 days. Comparable rankings are not
available for Apple iOS apps; SimilarWeb, Top Apps – Google Store/Top Free/Australia, 30 April 2021.
acma | 45
The most notable non-signatory is Snapchat. Snapchat is one of the most popular
mobile apps in Australia, with an estimated 6.4 million active monthly users 74, and a
high app usage rank. Against these measures, Snapchat appears to easily meet the
threshold of a ‘major’ platform and should be encouraged to participate in the code.
Pinterest and Reddit could also be considered major platforms by some metrics. Both
have high volumes of web-based traffic, each boasting over 6 million unique monthly
visits. Of the 2, Reddit is more commonly used as a source of general news and
information, and based on the N&MRC survey, is more widely regarded as a potential
source of misinformation. 75
While having less than 1% market share, privacy-oriented search engine DuckDuckGo
is the third most popular search engine in Australia, has experienced considerable
growth over the last 12 months, 76 and should also be encouraged to participate in the
code. Research out of the UK suggests that those who get ‘a great deal’ or ‘a fair
amount’ of COVID-19 information from DuckDuckGo have much higher levels of
vaccine hesitancy than the general population, and are also much more likely to
believe that reporters, scientists and government officials are involved in a conspiracy
to cover up important information about the coronavirus. 77
It should be reiterated that private messaging platforms remain outside the scope of
the code. As such, popular digital platform services like Facebook Messenger,
WhatsApp and WeChat are not covered, despite growing concerns that these
platforms are potential hotspots for misinformation. As discussed in Chapter 2,
WeChat has low overall usage within the community but high perceived volumes of
COVID-19 misinformation among its user base – second only to that of Facebook. 78
Alternative social media services

As also discussed in Chapter 2, smaller, encrypted private messaging, message board
and alternative social media platforms like Signal, Telegram, 4chan, 8kun, Gab and
Parler are also clear vectors for disinformation and misinformation content. These
services promote themselves as bastions of free speech and have minimal or less
restrictive content moderation policies, attracting conspiratorial communities that may
no longer be accepted on mainstream platforms. In particular, the use of Telegram and
Signal in Australia has increased in recent months, largely driven by concern over
WhatsApp policy changes and increased content moderation on other platforms. 79
Use of these services in Australia is currently too low for any to be considered a ‘major
platform’, and their respective stances on moderation make it highly unlikely that any
would consider signing up to a voluntary code of practice in Australia.
Nevertheless, governments, researchers and civil society should continue to actively

engage with these platforms on harmful misinformation narratives, and industry should
consider options for how to best engage with them as part of ongoing industry-wide
efforts to address disinformation and misinformation.
74
SocialMediaNews, Social Media Statistics Australia – April 2021, published 1 May 2021.
75
76
StatCounter, Search Engine Market Share Australia, May 2021.
77
University of Bristol and King’s College London, Coronavirus conspiracies and views of vaccination, 31
January 2021.
78
79 We Are Social, Social media insights into how online misinformation and disinformation are being spread
46 | acma
Finding 17: The bulk of ‘major platforms’ in Australia have signed up to the code. As
such, it should be regarded as an industry-wide initiative.
Finding 18: DIGI should continue to encourage other popular platforms, like Snapchat
and Reddit, to sign up to the code, even if they do not meet the proposed
threshold of 1 million active monthly users. DIGI should actively publicise
the involvement of any additional code signatories as soon as practicable
after their signing.
Finding 19: Industry participants should consider the role of private messaging
platforms and smaller alternative platforms in the amplification of
disinformation and misinformation and explore options for how these
platforms could be included within the code framework.
acma | 47
4. Assessment of the code
This chapter provides an overview of the code and an assessment of whether, in the
ACMA’s view, it has met the expectations set out by the government. Findings in this
chapter draw on submissions to the public consultation process, discussions with
DIGI, code signatories and other key stakeholders, and the ACMA’s expertise in code
development.
4.1. Regulatory approach and framework

The code takes an outcomes-based regulatory approach, specifying objectives and
outcomes that signatories commit to achieve. This section examines the general
approach taken in the code.
Table 7: Code objectives and outcomes
Objectives Outcomes
Objective 1: Provide safeguards Outcome 1a: Signatories contribute to reducing the risk of
against Harms that may arise from Harms that may arise from the propagation of
Disinformation and Misinformation. Disinformation and Misinformation on digital platforms by
adopting a range of scalable measures.
Outcome 1b: Users will be informed about the types of
behaviours and types of content that will be prohibited
and/or managed by Signatories under this Code.
Outcome 1c: Users can report content or behaviours to
Signatories that violates their policies … through publicly
available and accessible reporting tools.
Outcome 1d: Users will be able to access general
information about Signatories’ actions in response to
reports.
Objective 2: Disrupt advertising and Outcome 2: Advertising and/or monetisation incentives for
monetisation incentives for Disinformation are reduced.
Disinformation.
Objective 3: Work to ensure the Outcome 3: The risk that Inauthentic User Behaviours
integrity and security of services undermine the integrity and security of services and
and products delivered by digital products is reduced.
platforms.
Objective 4: Empower consumers Outcome 4: Users are enabled to make more informed
to make better informed choices of choices about the source of news and factual content
digital content. accessed via digital platforms and are better equipped to
identify Misinformation.
Objective 5: Improve public Outcome 5: Users are better informed about the source of
awareness of the source of Political Political Advertising.
Advertising carried on digital
platforms.
Objective 6: Strengthen public Outcome 6: Signatories support the efforts of independent

understanding of Disinformation researchers to improve public understanding of
and Misinformation through Disinformation and Misinformation.
support of strategic research.
Objective 7: Signatories publicise Outcome 7: The public can access information about the
the measures they take to combat measures Signatories have taken to combat Disinformation
Disinformation and Misinformation. and Misinformation.
48 | acma
ACMA commentary
It is extremely positive to see industry, steered by DIGI, come together to develop a
single code of practice. A single code should promote a consistent approach to
dealing with misinformation across platforms, while providing efficiencies through
standardised administration, complaints handling, and reporting processes. It also
means users only need to go to a single place to understand the protections offered by
signatories. This should promote confidence in industry to manage the range of harms
associated with misinformation.
The code is framed to address the Australian environment. While Australia or

Australians are not directly referenced in the code’s objectives and outcomes, the
definitions limit the scope of the code to digital content targeted at Australian users. 80
The effectiveness of the code in this area will depend on the extent to which
signatories tailor their measures for Australia, and provide relevant, local information
and data in their annual reporting on issues and measures affecting Australian users.
On issues of language, the code incorporates complex definitions and uses technical
jargon that may make aspects of the code unclear to users and the general public.
The development of definitions has been challenging due to the relatively novel nature
of the problem and lack of consensus on definitions among industry, researchers and
international organisations. 81 However, the ACMA considers that clear definitions and
simple language would make the code more accessible to the public and increase
transparency of platform measures.
The following provides a more detailed discussion on specific issues relating to the
regulatory approach and framework adopted by the code, including the guiding
principles, outcomes-based approach, and opt-in framework.
Guiding principles – protection of users’ rights

The code opens with a preamble that contextualises the problem and explains to
readers why the code was established. It also includes 7 guiding principles, which are
designed to inform the operation of the code and assist signatories in developing
suitable measures that do not, for example, impede on existing user protections or the
security of their services.
A key concern for government, articulated in its response to the DPI, is that the code
balances any interventions with rights to freedom of expression and speech. The code
acknowledges this issue upfront, with its first guiding principle focused on protection of
freedom of expression:
Protection of freedom of expression: Digital platforms provide a vital

avenue for the open exchange of opinion, speech, information,
research and debate and conversation as well as creative and other
expression across the Australian community. Signatories should not
be compelled by Governments or other parties to remove content
solely on the basis of its alleged falsity if the content would not
otherwise be unlawful. 82
This guiding principle also encourages signatories to be cognisant of the need to

protect internationally recognised human rights in developing proportionate responses
to disinformation and misinformation, including, but not limited to, freedom of speech.
Similar statements are also included in the EU Code (Appendix D).
80
Code provision 3.1.
81
Code provision 1.2.
82
Code provision 2.1.
acma | 49
The ACMA considers the preamble and statement of guiding principles a welcome
inclusion, particularly in flagging the need for signatories to protect user privacy,
support independent researchers, and balance protection from harms caused by the
propagation of disinformation and misinformation with freedom of expression and other
rights. It is important that no obligation is placed on signatories to remove content
merely on the basis of inaccuracy, where a threshold of serious harm is not met.
However, platforms still retain a responsibility to improve the online information

environment by implementing appropriate measures where a threshold of serious
harm is not met. Examples could include labelling of fact-checked content, introducing
friction to counter virality, or increasing transparency of information sources.
Outcomes-based approach
The ACMA considers the code’s outcomes-based approach is well suited to the nature
of the problem and the disparate business models and services of the major digital
platforms. Overall, the objectives and outcomes of the code are framed to provide
protections for both users and the general public. This is a key success of the code.
Under an outcomes approach, entities have the flexibility to develop their own
measures in a way that best reflects their services and business models. This allows
for innovation as industry adapts to a dynamic problem and makes progress towards
the achievement of common objectives.
The outcomes approach also provides platforms with the flexibility to implement
measures to counter disinformation and misinformation in proportion to the risk of
potential harm. This allows platforms to balance the need to address potential harms
with freedom of expression and other rights. Section 6 of the code details the need for
proportionality and the criteria that platforms may use when assessing the
appropriateness of their measures. Measures such as the removal of content or user
accounts can be appropriately limited to situations where there is a very high risk of
harm, while less stringent measures may be applied to lower-risk content in
accordance with platforms’ policies and procedures.
To be successful, an outcomes-based code depends on a high degree of commitment

from industry to work towards common, measurable outcomes established by the
code, and to demonstrate this through clear and robust performance reporting. More
information on an outcomes-based approach can be found in the ACMA’s position
paper.
Opt-in framework
A key feature of the code is its opt-in framework. All signatories must commit to
outcome 1(a) – reducing the risk of harms arising from disinformation and
misinformation – and to publish an annual report (see Table 8). Signatories are then
free to opt-in to the other objectives and outcomes. This approach is designed to
accommodate digital platforms that ‘operate vastly different businesses which offer a
wide and constantly evolving variety of services and products’ 83 and ‘the need of the
Signatories to choose those measures which are most suitable to address instances of
Disinformation and Misinformation’ on their services. 84 DIGI has noted that some
signatories may not have signed up to the code without the flexibility provided by the
opt-in model.
As noted above, the code has only 2 mandatory commitments. This provides a low
minimum standard for collective industry action, and by not specifying criteria for
83
Code provision 1.1.
84
See code provision 1.5.
50 | acma
opting out, potentially reduces transparency. Several submissions on the draft code
made similar criticisms of the opt-in model. 85
The code’s effectiveness will be dependent on widespread implementation of

measures to address disinformation and misinformation wherever such content exists.
While signatories’ reports show that they have by and large committed to all outcomes
relevant to their services (see Chapter 5), the ACMA considers that the code should
be strengthened by implementing an opt-out framework. Under such a framework,
platforms would be permitted to opt out of an outcome only where that outcome is not
relevant to their services. The recent European Commission (EC) guidance on the EU
Code suggests a similar approach (see Appendix D for broader discussion on
international approaches). 86
When opting out, platforms should be required to provide justification demonstrating

that the outcome is not relevant to their services. The ACMA considers that this
approach would provide signatories with sufficient flexibility to opt out where a service
is clearly not relevant, while providing greater transparency about signatories’
approaches and encouraging industry action. In addition, the proportionality principle
set out in Provision 6.1 means that a platform’s size, nature and available resources
should be taken into account when considering the appropriateness of a platform’s
measures, their ability to contribute to research, and initiatives under Outcomes 6 and
7 respectively.
As well as a high degree of industry commitment to take action to achieve agreed

outcomes, an outcomes-based model requires:
> concrete, measurable outcomes and a clearly defined scope that identifies and
directly targets the problem
> comprehensive reporting against agreed key performance indicators that measure
progress towards code outcomes
> a robust code-administration framework supported by a consistent program of
evaluation and review.
The code is assessed against these requirements in the following sections.
85
The Australian Associated Press noted the need for commitment to common objectives, stating that it is
‘essential that the common purpose is clearly identified in Australia’s code, and accepted by all signatories’
and that ‘expectations around engaging with the code to the fullest extent possible should be clearly stated’.
The ABC and SBS submissions observed that the opt-in model may reduce the incentive for platforms to
expand or improve their current initiatives. Reset Australia expressed scepticism that the opt-in model would
be effective in tackling disinformation. The Australian Muslim Advocacy Network argued that an opt-out
model would provide more scrutiny, with applications to opt out of an outcome to be made to an
independent administrator.
86
European Commission, European Commission, Guidance on Strengthening the Code of Practice on
Disinformation, 26 May 2021, p. 6.
acma | 51
Finding 20: DIGI has developed an outcomes-based code that has allowed
platforms with a range of business models to sign up to a single code.
Finding 21: The code objectives and principles meet the government objective of
striking a balance between encouraging platform interventions and
protecting freedom of expression, privacy and other rights.
Finding 22: The code should be strengthened by taking an opt-out approach. Opting
out of an outcome should be permitted only where the outcome is not
relevant to the signatory’s services. Signatories should provide
adequate justification when opting out.
4.2. Code scope

The scope of the code hinges on the definitions of disinformation (and inauthentic
behaviour), misinformation and harm. These definitions are outlined below.
Disinformation
The aspect of Disinformation that this Code focuses on is:
A. Digital Content that is verifiably false or misleading or deceptive;
B. is propagated amongst users of digital platforms via Inauthentic Behaviours; and
C. the dissemination of which is reasonably likely to cause Harm. [3.2]
Inauthentic Behaviour
Inauthentic behaviour includes spam and other forms of deceptive, manipulative or bulk,
aggressive behaviours (which may be perpetrated via automated systems) and includes
behaviours which are intended to artificially influence users’ online conversations and/or
to encourage users of digital platforms to propagate Digital Content. [3.5]
Misinformation
Misinformation means:
A. Digital Content (often legal) that is verifiably false or misleading or deceptive
B. is propagated by users of digital platforms; and
C. the dissemination of which is reasonably likely (but may not be clearly intended to)
cause Harm. [3.6]
Harm
Harm means harms which pose an imminent and serious threat to:
A. democratic political and policymaking processes such as voter fraud, voter

interference, voting misinformation; or
B. public goods such as the protection of citizens' health, protection of marginalised or

vulnerable groups, public safety and security or the environment. [3.4]
One of the key strengths of the code is that it covers both disinformation and
misinformation. This is a significant improvement on the consultation draft and
provides Australian users with stronger protections than those afforded under the
52 | acma
equivalent EU Code. 87 DIGI and code signatories should be applauded for listening to
stakeholder feedback on this issue and broadening the scope of the code.
The ACMA also welcomes the inclusion of a harm threshold. This acts as a
fundamental safeguard against platforms unnecessarily impinging on freedom of
expression. We are, however, concerned that the effectiveness of the code will be
limited by an excessively narrow definition, or interpretation, of harm.
The code does not require signatories to take action against content unless it is
reasonably likely to result in ‘serious’ and ‘imminent’ harm. The ACMA agrees that a
threshold of ‘serious’ harm is appropriate. This is in line with the DPI which
recommended a high threshold of ‘serious public detriment’ to avoid undue limits on
free expression. 88
However, the requirement that harm must also be ‘imminent’ introduces a temporal
element, which may be interpreted differently by signatories. If read narrowly, the
‘imminent’ test would likely exclude a range of chronic harms that can result from the
cumulative effect of misinformation over time, such as reductions in community
cohesion and a lessening of trust in public institutions. 89 As outlined in Chapter 2,
these types of chronic harms can increase vaccine hesitancy, promote disengagement
from democratic processes, and result in a range of tangible, real-world harms to both
individual users and society at large.
As it could have the effect of potentially excluding from the code a significant amount
of material of concern, the ACMA considers that the term ‘imminent’ should be
removed from the harm definition.
Services covered by the code

Section 4 of the code specifies the types of services and products that are covered by
the code.
Provision 4.1 articulates the scope of the code as applying to ‘services and products
that deliver to end users in Australia:
A. user-generated (including sponsored and shared) content; and/or
B. content that is returned and ranked by Search Engines in response to user
queries.
Provision 4.2 explicitly excludes the following services:

A. private messaging services including those provided via software applications;
B. email services including those provided via software applications;
C. Enterprise Services. 90
87
The EC guidance on the EU Code recommends that the code should be strengthened to commit
signatories to have in place policies and measures to mitigate the risks posed by misinformation where there
is a significant public harm dimension. European Commission, Guidance on Strengthening the Code of
Practice on Disinformation, 26 May 2021, p. 5.
88
Noting the DPI recommendation was in the context of a mandatory code addressing complaints about
content, where compliance would be assessed against whether code rules are followed in particular cases.
89
See ACMA, Misinformation and News Quality on Digital Platforms in Australia, p.12.
90
The code defines Enterprise Services as software and services including cloud storage and content
delivery services which are designed for the use of a specific organisation.
acma | 53
Provision 4.3 notes that the list of excluded services and products is not intended to
be exhaustive as new services and products are likely to emerge, some of which will
not be relevant to the code.
The code does not require signatories to nominate which of its services are covered.
Noting the open-ended nature of Provision 4.3, the ACMA considers that signatories
should be required to report annually on which services and products are covered by
the code. This is discussed further in Chapter 5.
Private and group messaging services

Under Provision 4.2(A), private messaging services are explicitly excluded from the
code. However, the code does not provide a definition of private messaging, which
creates some uncertainty about the range of services intended to be captured by this
exclusion.
Most private messaging services offer group message functionality, which can allow
for conversations between hundreds or even thousands of users at once. Some of
these larger groups take on the characteristics of a semi-public community or channel,
extending beyond friends and family to include strangers with shared interests. Our
environmental assessment in Chapter 2 indicates that the propagation of
disinformation and misinformation on these channels is increasingly problematic,
particularly on smaller, alternative platforms that allow for very large group sizes. 91
However, discussions with DIGI indicate that the intention is to exclude all messaging
products and services, including those that allow messaging to large groups. This is
justified on both privacy grounds, and due to technical limitations, such as end-to-end
encryption.
The ACMA agrees that private messaging services should be treated differently from
other digital platform services, and that there should be no requirement or expectation
that digital platforms should monitor private conversations between users. There are,
however, steps that platforms can – and in some cases, do – take to reduce the risk of
misinformation on messaging services, without needing to access or view the content.
Many of these measures are designed to reduce the amplification of content, for
example by introducing ‘friction’ that limits the speed or reach of forwarded messages
or shared links. As these services fall outside the scope of the code, platforms are not
required to be transparent about the nature and extent of the problem on messaging
services or about the effectiveness of their measures.
The ACMA’s position paper recommended that, given their role in spreading and
amplifying misinformation, online groups and semi-public channels such as one-to-
many or many-to-many messaging services should be covered by the code.
Stakeholders have also expressed concern about the exclusion of private messaging
services. In its submission to the draft code, the ABC noted the lack of clarity in this
exclusion, given the lack of a definition. The Digital Media Research Centre at
Queensland University of Technology suggested that private messaging be included
given the significant amount of disinformation spread through instant messaging. The
EC has also recently recommended the expansion of the EU Code to cover private
messaging services. 92
The ACMA considers there would be substantial public benefit in including messaging
services within the scope of the code, with the caveat that this should not entail any
91
Telegram, for example, allows messaging to groups of 200,000, and Signal to groups of 1,000.
92
European Commission, Guidance on Strengthening the Code of Practice on Disinformation, 26 May 2021,
p. 5.
54 | acma
obligation on platforms to monitor or censor the content of private messages. As
private, closed and semi-public groups or pages on social media platforms are already
included within the scope of user-generated content, 93 the inclusion of one-to-many
messaging services would support greater consistency across service types.
News aggregation services

It is also not clear whether news aggregation services are covered by the code. Both
the DPI and the ACMA’s position paper proposed that major platforms providing digital
content aggregation services should be included. Provision 4.1 appears to exclude
these services as they neither involve user-generated content nor are considered
search engines. DIGI has advised, however, that it considers these services to be
covered by the code, and Apple has signed up to the code on the basis of its Apple
News aggregation service.
News aggregators play a key role in disseminating news and information through the
online information ecosystem. Many stakeholders have noted the importance of quality
news being readily available online. If the code is to offer a comprehensive, industry-
wide approach to misinformation, disinformation and news quality, it is vital that news
aggregators delivering content to Australians are explicitly included within the scope of
the code. This should be clarified in later code revisions.
Content covered by the code

Section 4 of the code also places limitations on the types of content that is to be
covered by the code.
Provision 4.4 excludes the following types of content from the operation of the code,
unless signatories determine that specific instances of its propagation clearly fall
within the scope of disinformation:
A. content produced in good faith for entertainment (including satire and parody) or
for educational purposes;
B. content that is authorised by an Australian State or Federal Government;
C. subject to sections 5.21 to 5.23, Political Advertising or content authorised by a
political party registered under Australian law; and
D. news content that is the subject of a published editorial code which sets out
content standards and or/complaints mechanisms.
Provision 4.5 states that signatories may, at their discretion, ‘implement policies and
procedures which govern the dissemination by users on their platforms of the types
of content excluded from the operation of the provisions of the code under section
4.4, where signatories determine such content is reasonably likely to cause Harm.’
The ACMA considers the exclusion of entertainment and educational content

produced in good faith, government-approved content, and authorised political content
is appropriate given the need to balance interventions with freedom of expression. On
issues of professional news content and political advertising, however, greater clarity
would be welcomed.
Professional news content

Provision 4.4.D of the code excludes news subject to published editorial standards,
except in clear and specific instances of disinformation. 94 Although this wording seems
to exclude professional news from the code as a whole, DIGI has advised that this is
not the intention. Instead, the purpose is to clarify that platforms will not consider
93
Code provision 4.1.A.
94
Code provision 4.4.D.
acma | 55
professional news as misinformation under the code. That is, the code places no
obligation on platforms to extend counter-misinformation measures to professional
news content. 95
The ACMA acknowledges that professional news should be treated differently from
other types of online content, as most news content is already covered by separate
regulatory frameworks that promote accuracy and impartiality in reporting, and provide
separate avenues for complaints handling. However, there is concern that news
content that does not present a high risk of harm at the publisher level can sometimes
present a higher risk once it is taken out of that context. 96 For example, Chapter 2
outlines how links to news articles from reputable Australian media outlets are
commonly shared within conspiratorial communities to support misinformation
narratives. The code also leaves open the question of how news from sources not
subject to an Australian professional editorial code, such as non-Australian sources, is
to be treated.
In the ACMA’s view, the exclusion of professional news from misinformation is

unnecessary. The outcomes-based model can accommodate the different treatment of
professional news compared to other content, allowing platforms to apply different
measures in consideration of the editorial standards and complaints processes that
may already apply.
Political and issues-based advertising

The ACMA welcomes the objective to improve public awareness of the source of
political advertising (Objective 5). Political advertising is otherwise excluded from the
code, except where specific instances of it clearly fall within the scope of
disinformation. This refers to content that is propagated by inauthentic behaviours,
such as spam, bots, fake accounts or deceptive behaviours including foreign
interference and other information operations.
The ACMA considers this exclusion to be appropriate. Political advertising is covered

by existing electoral law obligations, and the protection of freedom of political
expression is critical.
However, it is unclear whether issues-based advertising is also excluded from the

code. 97 Issues-based advertising includes sponsored and paid-for content that is
intended to bring awareness to, advocate for, or call for action on certain topics that
are widely discussed in the public sphere, such as political and social issues.
Issues-based advertising is a known vector of misinformation. There is particular

concern about the ability of micro-targeting technologies, which rely on user data, to
direct advertisements containing false or misleading information at particular groups
and not others. Micro-targeting can reduce transparency, as advertisements are visible
only to those who are targeted.
To alleviate these concerns, the code should include a clear definition of issues-based
advertising and the scope of Objective 5 should be extended to include issues-based
advertising. Providing users with greater transparency about the source of issues-
95
Email from DIGI ‘Additional ACMA questions’, received 15 June 2021.
96
For example, corrections made by the publisher may not be clear to users once the original article has
been shared; publishers no longer have control over how the content is used; the content may be
manipulated; or removal of original context may make the content misleading.
97
The definition of political advertising in the code includes advertisements that ‘advocate for the outcome of
a political campaign concerning a social issue of public concern in Australia.’ The meaning of ‘advocate for
the outcome of a political campaign’ is not defined.
56 | acma
based advertising would increase users’ awareness of why they are being targeted
and inform their decision-making about important social issues.
Several platforms currently have separate policies on issues-based advertising.

Explicitly including issues-based advertising within the scope of the code would
increase transparency of their measures and encourage other platforms to implement
similar measures.
There are concerns that placing limits on issues-based advertising would unduly limit
political expression. Acknowledging this, the EU code commits signatories to
improving the transparency of issue-based advertising, and tasks them with the
development of a working definition of issue-based advertising which does not limit
reporting on political discussion and the publishing of political opinion. The ACMA
recommends a similar approach be taken in the Australian code.
Paid and sponsored content

The ACMA considers that platforms have a greater level of responsibility for paid and
sponsored content, given their greater control over the content and the monetary
benefit they receive. The potential to target content at individuals or groups also
increases the potential for harm caused by disinformation and misinformation
propagated through advertising channels. Signatories should take a more active role,
and implement more proactive measures, in relation to this type of content.
The treatment of paid and sponsored content should be made clearer in the code.
While sponsored content is explicitly included within the scope of user-generated
content, the term is not defined in the code. Improved clarity around the operation of
these arrangements would be beneficial for platforms, advertisers, and users.
Finding 23: The code covers both disinformation and misinformation. This is one of
the key strengths of the code, and is an improvement on the current EU
Code.
Finding 24: The definition of harm in the code is too narrow to provide adequate
safeguards against the full range of harms caused by the propagation of
disinformation and misinformation.
Finding 25: Private messaging services should be included within the scope of the
code as these are known vectors of disinformation and misinformation.
These should be included with appropriate caveats on the right to
privacy.
Finding 26: The code should clarify that the exclusion of professional news content
applies only to the application of counter-misinformation measures. It
should also clarify that news aggregation services are in scope.
Finding 27: The treatment of paid and sponsored content should be clearer in the
code. This should include a clear definition of sponsored content.
Finding 28: In addition to improving public awareness of the source of political

advertising, the code should also cover the source of issues-based
advertising.
acma | 57
4.3. Code objectives, outcomes and commitments
There is room for improvement in the drafting of the code. Some outcomes could be
considered outputs, and this may encourage reporting focused on providing data to
quantify measures implemented by platforms, rather than data on the effectiveness of
those measures. Developing and embedding metrics in the reporting framework
currently under development may address these concerns.
A positive element in the code is the inclusion of example measures under each code
outcome covering a wide range of approaches. However, the code would benefit from
more detailed commitments that lay a foundation for the development of consistent,
evidence-based measures across the industry. Examples are discussed under the
relevant sections below.
Safeguards against disinformation and misinformation

The list of example measures at 5.9 includes a wide range of both proactive and
enabling measures to reduce the impact of disinformation and misinformation.
Importantly, it includes the adjustment of ranking algorithms to reduce user exposure.
More-detailed, structured commitments to address disinformation and misinformation
might include:
> consideration of a common framework for the assessment of harm that can help
structure platforms’ internal decision-making
> the establishment of mechanisms for the exchange of information between
platforms on emerging disinformation and misinformation risks
> concrete commitments to mitigate risks of recommender systems fuelling the viral
spread of disinformation, as recently recommended in the EC’s guidance for
strengthening the EU code. 98
Disrupting advertising and monetisation incentives for disinformation

The provisions under Objective 2 are focused on providing tools and information to
advertisers, as well as restricting advertising availability on accounts and websites that
propagate disinformation. The ACMA is concerned that these provisions focus
exclusively on providing such tools and not on the responsibility of platforms to
scrutinise advertisements placed via their systems.
These provisions are weaker than those in the EU code, which state that signatories
will use commercially reasonable efforts not to accept remuneration from, or otherwise
promote accounts and websites that consistently misrepresent information about
themselves. The EC’s guidance on reforming the EU code seeks to strengthen these
provisions. In particular, it states that platforms should commit to ‘tighten eligibility
requirements and content review processes for content monetisation and ad revenue
share programmes on their services’. 99
The code could also establish cooperative structures for exchanging information on
sources of monetised disinformation. This could include a common repository of
rejected advertisements, as noted in the EC’s Guidance on Strengthening the EU
Code of Practice.
98
p. 14.
99
p. 7.
58 | acma
Empowering users to identify the quality of news and information
The code includes a range of both proactive and enabling measures to allow users to
make more informed choices of news and factual content.
Importantly, it includes broader measures that go beyond the immediate digital

platform environment, such as the promotion of digital literacy and the provision of
financial support or sustainable partnerships with fact-checking organisations. Several
stakeholders have noted the importance of developing sustainable partnerships.
More-detailed commitments to improve the online information environment could

include measures that address the role that system design can play in the propagation
of disinformation and misinformation, and a commitment to increased transparency
over the criteria that platforms (or their algorithms) use to assess the quality of
information and to prioritise content.
Many stakeholders have expressed the need for the code to address algorithmic
transparency, and this was a concern raised in the DPI. 100 The EC guidance on reform
of the EU code also calls for increased transparency in the criteria used for prioritising
or de-prioritising information, with the option for users to customise ranking algorithms.
Other, more-structured measures could include frameworks to establish collaboration

with experts in media literacy and safety by design, and with fact-checking and news-
industry bodies. These would provide channels for expert advice and include
agreements to collaborate on the development and implementation of tools and
initiatives to improve users’ ability to navigate the online information environment.
Several stakeholders have noted that there is inadequate recognition in the code of
the importance of working closely with news industry bodies or news media.
Collaboration with researchers, government and other stakeholders

The code includes a guiding principle recognising the importance of industry support
for independent research. 101 Provisions for strengthening the public understanding of
disinformation and misinformation through the support of strategic research are also
included under Objective 6. Stakeholders have also observed that a similar outcome is
included in the EU Code but that progress on this outcome has been slow.
As noted by several stakeholders, more detailed and structured commitments would

be valuable in promoting improvement and transparency in this area. These could
include frameworks to establish ongoing collaboration with researchers, government
and other stakeholders, and could address such matters as data-sharing
arrangements to facilitate research, real-time monitoring of disinformation and
misinformation or the formation of expert advisory bodies on matters of concern to the
public, government, or other industries.
100
See submissions on the draft code by Australian Strategic Policy Institute, Centre for Responsible
Technology, Reset Australia, Digital Rights Watch, the Australian Muslim Action Network, RMIT ABC Fact
Check and SBS.
101
Code provision 2.6.
acma | 59
Finding 29: The output-focused framing of several objectives and measures may
encourage reporting to focus on outputs rather than progress towards
code outcomes.
Finding 30: The code should include industry-wide frameworks for the development
and implementation of individual platform measures. Examples could
include frameworks to establish:
> criteria for the assessment of harm
> criteria for assessing news and information quality
> processes for the exchange of information between platforms on
disinformation and misinformation risks
> commitments to address the propagation of disinformation and
misinformation via platform advertising channels
> commitments to address the risks of propagation via platform
algorithms and architecture.
4.4. Code administration and reporting regime

DIGI will be the administrator of the code and will establish a sub-committee to meet at
6-monthly intervals to review the actions of the signatories and monitor how they are
meeting their commitments under the code. 102 The sub-committee will include
independent members as well as signatory representatives.
This is an important measure and a sign that the industry is prepared to move towards
greater public transparency and accountability. However, no detail is provided in the
code on how the sub-committee will operate, including how the independent members
will be selected, or the circumstances which might lead a signatory’s actions to be
considered by the sub-committee.
There is also very little detail on enforcement mechanisms. Provision 7.4 states that
signatories agree to develop and document a process describing circumstances in
which a non-compliant signatory may be removed. This is an important consideration
but removing a signatory may impact the effectiveness of the code as a self-regulatory
mechanism. A range of enforcement mechanisms beyond removal should be
considered.
The lack of detail on code administration limits the ACMA’s ability to assess the likely
effectiveness of the code. The ACMA considers that the inclusion in the code itself of a
framework setting out principles for the structure and operation of the sub-committee
would provide greater transparency and accountability.
Code reviews
An initial code review will take place after 12 months of operation. 103 Subsequent
reviews will occur at 2-yearly intervals. These reviews will be based on the input of the
signatories, relevant government bodies (including the ACMA) and other interested
stakeholders, including academics and representatives from civil society active in this
field.
The ACMA considers the reviews are appropriately timed. The initial review will allow
feedback to be considered on the content and structure of the code, the adequacy of
platform reporting, the extent of signatories’ commitments to the code and the
102
Code provision 7.5.
103
Code provision 7.6.
60 | acma
development of code administration processes. This includes the opportunity to
incorporate findings from this report and for other developments to be considered.
The reviews would benefit from the input of a broad range of stakeholders. They
should include a public consultation process that is promoted through appropriate
channels including on signatories’ digital platform services. They should also take into
account the input of stakeholders from related industries, including news media and
health.
Code complaints mechanism

The code includes a commitment to establish a facility for addressing code non-
compliance, including a mechanism for handling unresolved complaints. 104 The
complaints facility will be established within 6 months of the commencement of the
code. The facility will hear appeals of complaints of code breaches that have not been
acted upon by signatories, but not individual complaints of signatories’ decisions
regarding content on their platforms, including whether specific items of content should
be retained or removed.
The ACMA considers this to be suitable approach for addressing general complaints
about code compliance. However, the ACMA is concerned that the code does not
place obligations on signatories to have robust internal complaints processes to
address user complaints. The ACMA expressed this expectation in its position
paper. 105
The code should set an expectation that signatories will have an internal complaints
process for matters covered by the code that is transparent, responsive and
accessible. Complainants should access this internal process in the first instance. If
they are unable to resolve the complaint internally, signatories should provide access
to external dispute resolution, so that the matter can be considered and resolved by an
independent third party at no cost to the complainant.
The government is committed to considering an external dispute resolution scheme as

part of its response to the DPI (outlined in Appendix E). Platforms may wish to
consider whether this is an appropriate forum to address escalated complaints under
the code.
The code should also set out how complaints, including escalated individual
complaints, may be referred to the complaints facility established under the code. It is
important to note that there is a point at which individual signatory decisions about
content may become a matter of code compliance. For example, where a platform’s
decision on a piece of content, or the process it follows in making that decision, do not
comply with the platform’s published policies, or where the policies themselves do not
provide adequate protections in accordance with the code.
Performance reporting framework

Signatories are required to submit annual reports that set out progress towards
achieving code outcomes and will be published on the DIGI website. 106
Appendix 2 to the code provides a template for an initial report that signatories are
required to submit within 3 months of signing up to the code. This template provides a
workable foundation for platform reporting. Key areas where the reporting template
could be improved include:
104
Code provision 7.4.
105
ACMA, Misinformation and News Quality on Digital Platforms in Australia, p. 25.
106
Code provision 7.3.
acma | 61
> a clear format to set up existing measures, proposed measures, and performance
reporting separately under each outcome
> a clear requirement for performance reporting to provide adequate data to
measure platform performance against each outcome and not just describe the
actions platforms have taken. For example, outcome 1(c) in Appendix 2 currently
only asks that signatories include links to published policies, procedures and
guidelines
> a more detailed discussion about future plans against each outcome in the report,
which would provide greater visibility about changes over time
> a clearer distinction between the identification of relevant measures or actions
(that is, what steps are individual signatories committing to do under the code) and
ongoing reporting on the effectiveness of these measures in addressing the code
outcomes (for example, how individual signatories intend to demonstrate their
measures or actions have been successful).
A more detailed discussion on the application of the template is contained in

Chapter 5.
Within 6 months of code commencement, signatories will develop and implement an

agreed format for annual reports and a guideline that will inform the data and other
information to be included in those reports.
The development of a robust reporting framework is critical to the effectiveness of the

code. This development would benefit greatly from broad input and collaboration,
including with academic experts and stakeholders in related industries. The guideline
should also include the development of standard key performance indicators against
each outcome. These are crucial to improving the transparency of signatory actions,
encouraging industry progress towards code objectives and monitoring the code over
time.
While the ACMA acknowledges the commitment to develop a guideline, without one it
is difficult for us to provide a full assessment of the code’s reporting framework. The
ACMA would welcome the opportunity to provide input to signatories on their proposed
guidelines and reporting framework, including collaborating on identifying key metrics
and indicators. Some initial guidance is provided at Appendix F.
Finding 31: The code provides a high-level code administration framework. Given
that detailed arrangements for code administration, compliance with the
code, and consumer complaints are still under development, the
ACMA’s ability to assess their practical effectiveness is constrained.
Finding 32: The code should include a framework setting out principles for the
structure and operation of the sub-committee to provide greater
transparency and accountability.
62 | acma
Finding 33: The reporting template provides a workable foundation for the reporting
guideline. Reporting should incorporate adequate data to measures
performance against KPIs under each outcome; detailed action plans to
address areas identified for improvement; and a clearer distinction
between measures (that is, outputs) and the effectiveness of these
measures (progress towards outcomes).
Finding 34: The proposed 12-month review will provide an opportunity for findings
from this report, and other developments, to be incorporated into the
code.
Finding 35: The lack of detail on code administration matters, including on the
operation of the sub-committee and guidelines for future code reporting,
has limited the ACMA’s ability to undertake a full assessment on the
likely effectiveness of the code.
acma | 63
5. Assessment of platform
performance
As a mandatory commitment of the code, each signatory is required to provide an
annual report to DIGI setting out its progress towards achieving the code objectives it
has opted-in to.
While the code has not been operational for long, the ACMA has met with all code
signatories, reviewed their initial annual reports, and is able to draw some
preliminary observations on performance based on this information.
This chapter provides a thematic analysis of signatories’ initial annual reports,

covering:
> how signatories have met their reporting obligations under the code
> the range of measures signatories have identified to meet their commitments
> the effectiveness of these measures (noting the short time the code has been in
place and limited amount of data provided by signatories)
> general reporting requirements and code commitments.
Appendix B includes individual assessments of signatories’ reports.
5.1. Platform commitments

The first stage of the ACMA’s assessment was to examine platform commitments and
whether signatories had met their general reporting requirements under the code. The
code prescribes a reporting template, asking signatories to provide information on
their business, measures against the relevant outcomes the platform has opted-in
to, approach to monitoring performance, and information about future trends.
Opt-in nomination forms and initial annual reports by signatories were published on
DIGI’s website on 22 May 2021, 3 months after the commencement of the code.
An outline of signatories’ opt-in commitments is provided in Table 8 below.
It should be noted that the table is included for clarity only. On the whole, the ACMA
considers that where signatories have chosen to opt out of particular outcomes they
have done so on justifiable grounds. Some comments on particular decisions are
included in individual platform assessments in Appendix B.
64 | acma
Table 8: List of signatories’ code commitments
Outcome Adobe Apple Facebook Google Microsoft Redbubble TikTok Twitter
(Mandatory) 1a: Signatories contribute to reducing the risk of harms that may
arise from the propagation of disinformation and misinformation on digital ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
platforms by adopting a range of scalable measures.
1b: Users will be informed about the types of behaviours and types of content
that will be prohibited and/or managed by signatories under this code. ✓ ✓ ✓ ✓ ✓ ✓
1c: Users can report content and behaviours to signatories that violates their
policies under 5.10 through publicly available and accessible reporting tools. ✓ ✓ ✓ ✓ ✓ ✓
1d: Users will be able to access general information about signatories’ actions
in response to reports made under 5.11. ✓ ✓ ✓ ✓ ✓ ✓
2: Advertising and/or monetisation incentives for disinformation are reduced.

✓ ✓ ✓ ✓ ✓ 107 ✓ ✓
3: The risk that Inauthentic User Behaviours undermine the integrity and
security of services and products is reduced. ✓ ✓ ✓ ✓ ✓ ✓
4: Users are enabled to make more informed choices about the source of
news and factual content accessed via digital platforms and are better ✓ ✓ ✓ ✓ ✓ ✓ ✓
equipped to identify misinformation.
5: Users are better informed about the source of political advertising.
✓ ✓ ✓ ✓
6: Signatories support the efforts of independent researchers to improve

public understanding of disinformation and misinformation. ✓ ✓ ✓ ✓ 108 ✓ ✓ ✓
(Mandatory) 7: The public can access information about the measures

Signatories have taken to combat disinformation and misinformation. ✓ ✓ 109 ✓ ✓ ✓ ✓ ✓ ✓
107
Redbubble has opted in to implement policies and processes that aim to disrupt advertising and/or monetisation incentives for disinformation (5.14) but has opted out of committing to
examples under 5.14 as no third-party advertising is permitted to be published on Redbubble. Redbubble does not sell media space to any third-party businesses (5.15, 5.16).
108
Microsoft has not yet committed to opt in to convening an annual event to foster discussions regarding disinformation within academia and civil society (5.27).
109
Apple will make and publish an annual report, as a requirement under the code (5.28), but initially will not publish additional information detailing their progress or additional commitments they
have made under the code, such as additional reports or public updates (5.29, 5.30).
acma | 65
ACMA commentary
All of the signatories met the stipulated timeframe outlined in the code, providing DIGI
with their opt-in nomination forms and interim annual reports within the 3-month
deadline. While Adobe and Apple were not initial signatories, it is pleasing to see that
they were able to meet this timeframe.
Most of these reports contain information against each of the categories stipulated in
the template. However, signatories have taken varied approaches to structuring and
formatting their reports. Future reporting would benefit from an agreed, uniform
approach across all signatories, to aid direct comparison.
A key focus for the ACMA was on whether signatories had fully articulated their
rationale for not opting into specific code outcomes. For the most part, we considered
these explanations to be relatively clear; however, in some cases they were not
sufficiently targeted to the outcome. For example, Apple did not state why it chose not
to opt-in to providing additional information under Objective 7, and Adobe provided a
general justification that did not explain why it chose not to opt-in to seemingly relevant
outcomes, like supporting strategic research under Objective 6. By contrast,
Redbubble offered clear justifications for its decisions, including that it does not
disseminate news or permit advertising, and that it is a small company with limited
resources for reporting additional information outside of its annual report.
Most signatories were also clear about which of their products and services would be
covered by the code. However, there were instances where reports refer to secondary
products and services, such as ad service technologies and enterprise software, and it
was not always clear why they were included or whether signatories intended for them
to be covered. For example, Microsoft reported that all its consumer-facing services
will be covered without providing a specific list of services, Facebook included
information about measures on excluded services, and Adobe’s commitments related
to the use of a technology rather than a specific product or service.
The ACMA considers that signatories should be required, in future annual reporting, to
provide a list of which services and products are covered by the code. Signatories
should also justify the exclusion of any major product or service offering on the basis
of specified criteria. Where appropriate, this justification should include the provision of
data demonstrating that there is no serious risk of harm associated with the
propagation of disinformation or misinformation on that product or service.
Finding 36: On the whole, signatories have met the initial reporting requirements set
out in the code.
Finding 37: For the most part, signatories have provided appropriate explanations
where they have not opted-in to specific commitments.
Finding 38: A more uniform approach to reporting would assist in cross-platform

assessment and increase transparency of platform measures and
performance.
Finding 39: For future reports, signatories should clearly specify the products and
services covered by the code, and justify any major exclusions.
66 | acma
5.2. Suitability of platform measures
The next stage of the ACMA’s assessment was to evaluate the appropriateness or
suitability of the measures identified by signatories. This included consideration of
whether the measures are appropriately targeted to their relevant code outcome(s),
including a clear nexus to Australia or Australian users, and a commitment to continual
improvement and future initiatives. We also examined whether key concepts in the
code had been applied in a consistent manner across signatories.
In assessing the suitability of these measures, we must first recognise the dynamic
nature of disinformation and misinformation. Over the last 18 months, the number of
platform measures has expanded across industry, largely in response to the COVID-
19 pandemic. This demonstrates a responsiveness to changes in the information
environment, but also raises questions as to whether these efforts are temporary or
signal stronger industry-wide action to address the challenges of disinformation and
misinformation.
While this discussion is primarily based on the information contained within signatory
reports, the ACMA has also separately been monitoring changes in platform policies
over the last 18 months, as reflected in the timeline at Appendix C.
ACMA commentary
Interim annual reports show that signatories have adopted a wide range of measures
to address the problems of disinformation and misinformation and to improve the
quality of news and information on their services.
Most of the measures identified in the initial reports are pre-existing and based on
global, platform-wide policies and community standards, rather than new measures
driven by the introduction of the code or that expressly target Australian users. This is
not unexpected, particularly as the code is a new initiative, and Australia is a relatively
small market.
Notwithstanding this, some signatories did identify measures they have implemented,
or are planning to implement, that focus on Australia:
> signatories including Facebook, Google and TikTok discussed their partnerships
with local fact-checking organisations
> several signatories, including Apple, Facebook, Google, Twitter and TikTok,
provided information on measures they had implemented in response to events
directly relevant to Australia, including dedicated spaces to prioritise quality news
and locally sourced authoritative information relating to high-risk matters including
COVID-19 and natural disasters.
Future initiatives
One disappointing aspect of the reports was the limited discussion about future
measures or initiatives that signatories are planning to introduce under the code. In
general, the reports heavily focused on current measures and past actions, and
signatories mostly avoided providing any concrete information on planned
developments or future initiatives. The one exception was Facebook, which included
information on its plans to expand its policies to improve transparency of the source of
political advertising to issues-based advertising in Australia.
While we recognise the challenges in providing industry and bad actors with advance
notice of changes, it is important for signatories to signal any broad commitments that
are planned to address misinformation. This would strengthen future reporting and
better enable stakeholders to track the rollout of new measures under the code.
acma | 67
Interpretation and use of code terms
The reports show inconsistencies in the interpretation and use of key code terms
between signatories. In discussing measures, several signatories referenced
definitions of ‘disinformation’ and ‘misinformation’ from their internal policies and
community standards, rather than the agreed definitions under the code. This presents
a fundamental issue in the construction of the code, hampering assessment of
signatories’ performance against the code and any industry-wide comparisons.
Similar issues have been raised within the context of the EU code, with recent
European Commission guidance recommending a harmonised template that allows, to
the extent practicable, cross-platform comparisons.
Some examples of this include:

> Facebook used different definitions of disinformation and misinformation from the
code, with disinformation being used to refer to inauthentic behaviour with the
intention to deceive, and misinformation to refer to content that is false or
misleading. This reflects Facebook’s existing policies and the different tools it uses
to address these 2 problems. Despite this, Facebook’s measures appear to cover
the scope of the code.
> TikTok used a narrower definition of misinformation than in the code, defining it as
false or inaccurate content. Content that is misleading is considered only in the
context of elections and civic processes, and manipulated media. Disinformation
was included as coordinated inauthentic behaviour to exert influence and sway
public opinion while misleading individuals and our community about the account's
identity, location, or purpose, and was included under misinformation.
> Google noted that, in practice, it does not make a distinction between
disinformation and misinformation in the application of its policies.
The reports also included detail of content and services that are excluded from the
scope of the code. For example, Facebook provided some information on its private
messaging service, Facebook Messenger, despite these types of services being
excluded from the code. The additional information is welcome to provide visibility
about signatory actions about misinformation. However, a revised reporting template
should make it clear that the provision of this information relates to excluded content
and services.
Assessment of harm and proportionality

The way in which signatories assess harm and apply proportionality and risk
considerations under their policies is not always transparent in their reports. This could
be clearer in future reports. For example, TikTok’s misinformation policy does not limit
harms to those who pose an imminent and serious threat. However, its approach to
assessing the extent of harm, and therefore whether content should be removed or
labelled, is not clearly articulated.
Other signatories also reported their approach to harms below the serious and
imminent threshold set out in the code. The ACMA welcomes this additional reporting
given we consider this threshold to be too high to capture the full scope of potential
chronic and acute harms (see Chapter 4). Both Google and Facebook set out
graduated approaches to disinformation and misinformation based on the risk of harm.
Facebook referred to its COVID-19 misinformation policies, which are transparent
about particular topics and claims that it considers to be harmful. It noted that these
are established in collaboration with experts in health communication and other fields.
Facebook also referred to its policy of reducing content that does not contravene its
community standards but has been rated as false by fact-checkers.
68 | acma
User reporting of disinformation and misinformation to platforms
The code sets the expectation that signatories will have functions for users to report all
disinformation and misinformation content. 110 All signatories who opted into this
provision provided descriptions of how users can report content that contravenes their
policies. Google, Microsoft and Redbubble also provided screenshots of their existing
reporting functions. In all cases, signatories provide tools to allow users to report
content via their platforms.
However, not all signatories allow users to report content under all of their
disinformation and misinformation policies or for all relevant aspects of their services.
For example, Twitter users are only able to report content under its platform
manipulation and spam policy, but not under its COVID-19 misleading information
policy. Facebook allows users to report content on Facebook and Instagram as ‘false
information’, which will be assessed under its misinformation policies. It does not
consider that users would be able to detect inauthentic behaviour, and as such does
not provide reporting mechanisms for suspected disinformation.
The code also sets the expectation that signatories will publish general transparency
data on the actions taken in response to user reports and the reporting template asks
platforms to provide information on how they do this. 111
In most cases, signatories reported that they provide general information about their
responses to user complaints in their public transparency reports. Transparency
reports provide aggregated data on signatories’ enforcement of their policies and, in
some cases, provide information on user complaints data. For example, YouTube
reports on videos removed by source of first detection – automated flagging, user,
trusted flagger, NGO or government agency – as well as on videos removed by
removal reason – harmful or dangerous, harassment, hateful or abusive, violence and
violent extremism and other. However, information of this kind as it relates specifically
to the code was not included in any signatories’ reports.
On the whole, signatories provided limited information on their procedures for

processing reports from users, including whether users could expect to receive a
response. Twitter reported that users would receive a response directly from their
support teams about the results of any investigation or enforcement action. TikTok
reported that creators are notified when their content has been found to violate its
policies but did not comment on what information is received by users who make a
complaint. Although not included in their initial report, YouTube users are able to view
their reporting history and the action that was taken by YouTube in response to the
report. Redbubble told the ACMA that it provided an automated message in response
to user reports, but that it would not contact the user directly unless it needed more
information, or if the user had reported the content via social media.
From discussions with signatories, it is clear that in many circumstances, the onus will
fall on the user to check whether the platform has taken action against reported
content, rather than receiving direct notification. This lack of information is consistent
with the DPI finding that there is substantial room for improvement in digital platforms’
user reporting and internal dispute resolution processes, including by increasing
transparency and consumer access. 112 Signatories should look to increase
transparency by improving the information they provide to users about the outcomes of
their reports.
110
Code provision 5.11.
111
Code provision 5.13.
112
ACCC, Digital Platforms Inquiry Final Report, 2019, pp. 507–9.
acma | 69
Need for further clarity and specificity
On the whole, signatories provided concrete and detailed descriptions of their
measures under each outcome. In some cases, however, it was not completely clear
how certain measures will contribute to the achievement of the code outcomes under
which they have been reported.
For example, TikTok reported its Asia–Pacific Safety Advisory Council as a measure
against outcome 6, which aims to improve public understanding of disinformation and
misinformation. It is not clear, however, whether the council's findings or advice will be
made publicly available, thereby improving public understanding of disinformation and
misinformation. Outside its report, TikTok has advised that the committee will provide
advice to TikTok on its content moderation policies, which suggests it is a measure
targeted at Objective 1.
The case study below provides a high-level assessment of the reported measures
against Objective 4. Signatories reported a range of measures that appear to be
relevant to achieving the objective. However, signatories’ reporting was not detailed, in
terms of either quantitative or qualitative data, making it difficult to assess the
relevance and impact of measures on users. This is a common theme across the
reports provided.
70 | acma
Case study: Assessment of reported measures against Objective 4
Objective 4: Empower consumers to make better informed choices of digital content.
Signatories reported a broad range of measures intended to enable users to make more-
informed choices about the news and factual content they access via digital platforms and
to identify misinformation.
Facebook, Twitter, TikTok and Google reported on their financial support for media literacy
programs.
Google, Microsoft and Facebook outlined measures to prioritise authoritative information in

search results. For example, Google and Microsoft reported making ongoing
improvements to promote authoritative sources and demoting borderline or low-authority
sources in Google Search and Bing. Microsoft also reported on a specific feature,
Intelligent Search, which is designed to promote a diversity of perspectives by displaying
‘all valid answers to a question’ in a carousel.
Apple, Microsoft and Google reported on their news aggregation services, which provide
news content subject to an independent editorial code and complaints scheme. Google’s
Full Coverage feature in Google News also provides videos and articles from different
publishers on a news story selected by the user. Apple’s and Microsoft’s services both
involve human editorial input.
Many signatories reported employing technologies to signal the credibility of news

sources. Facebook, Google and Twitter label government-controlled channels and
accounts, and Apple maintains brand and logo information on individual articles in Apple
News. Signatories also outlined measures to assist users or platforms to check the
accuracy of online news content or identify its provenance. For example:
> Apple News and Bing News both partner with NewsGuard, a news-rating organisation
> Facebook, Google, Microsoft, Twitter and TikTok all partner with fact-checkers and
apply labels to fact-checked information
> Adobe is developing a system for creators and publishers to embed visual and audio-
visual content with attribution data as part of its CAI.
A number of measures reported by signatories against Objective 1 are likely to contribute

to helping users identify misinformation and therefore are relevant to Objective 4. These
include prompts to authoritative sources when searching for significant social topics, such
as COVID-19. Facebook also reported a feature to direct users to authoritative sources if
they have previously encountered COVID-19-related content subsequently identified as
false by fact-checkers.
Across their reports, signatories have reported a range of measures that appear to be
relevant to achieving Objective 4. However, signatories’ reporting is not detailed, in terms
of either quantitative or qualitative data, making it difficult to assess the relevance and
impact of measures on users.
Broadly, we observe that signatories could strengthen commitments by implementing

further measures in collaboration with news providers. For example, credibility signalling
and fact-checking initiatives would be enhanced by a more holistic approach. This was
also raised by stakeholders who considered that news quality was a primary mechanism
for countering misinformation. Stakeholders also suggested that credibility signalling of
news sources could be enhanced by displaying logos of relevant journalistic or industry
bodies.
acma | 71
Finding 40: Signatories have a wide range of measures in place to address the
problems of disinformation and misinformation and to improve the
quality of news and information on their services. They also
demonstrate responsiveness to significant changes over the last 18
months, as well as to public and government calls for stronger action.
Finding 41: It is expected that signatories will develop more Australia-focused

measures over time.
Finding 42: In general, the reports are heavily focused on current measures and
past actions, and signatories have provided little systematic information
on future initiatives. In some cases, it is not clear to what extent certain
measures will contribute to the achievement of the code outcomes
under which they have been reported.
Finding 43: There are inconsistencies in the interpretations of key terms between
signatories, which are drawn from pre-existing definitions from their
internal, often global, policies. This makes it difficult to interpret and
assess performance and to make industry-wide comparisons.
Finding 44: A harmonised template would assist in comparing initiatives across

platforms. It would also allow clear reporting on additional information
beyond the requirements of the code.
Finding 45: Signatories have provided a large range of information on the actions
they have taken to address misinformation, disinformation and news
quality and to invest in collaborative initiatives. This demonstrates
signatories’ commitment to addressing these issues.
Finding 46: The information signatories have provided is heavily focused on platform
outputs and on volumetric data. Reporting lacks systematic data,
metrics or key performance indicators (KPIs) that establish a baseline
and enable the tracking of platform and industry performance against
code outcomes over time.
5.3. Effectiveness of platform measures

An important component of an outcomes-based approach is the ability of entities to
demonstrate the effectiveness of the measures they have adopted to meet the
identified outcomes. It follows that entities must be able to provide sufficient
information and data about their respective measures in order to baseline, monitor and
track their performance against the stated outcomes.
While the code has not been in operation for long, the ACMA has considered some
preliminary findings on the effectiveness of signatories’ measures against their
respective objectives.
Measures to address misinformation and news quality

Reports provided a large range of information and data on the actions that signatories
have taken to address misinformation, disinformation, and news quality. This gives an
indication of the nature and extent of these issues on their services. It also
demonstrates the signatories’ commitment to addressing these issues via a variety of
measures and initiatives:
72 | acma
> Google, Facebook, Twitter and TikTok provided information on their transparency
centres and transparency reports, which collate information and data on the
actions they are taking to address online safety and authenticity.
> Google and Facebook provided recent global data on the number of accounts
removed for engaging in coordinated inauthentic behaviour and influence
operations.
> Facebook, Google and TikTok provided data on the number of takedowns of
potentially harmful misinformation relating to COVID-19, including Australia-
specific data.
> Google provided a case study on its approach to COVID-19 misinformation,
including raw data on impressions of authoritative information panels.
Several signatories provided data on their investment in measures and initiatives to

combat disinformation and misinformation and improve the quality of online news and
information:
> Google provided expenditure data relating to the Ad Grants Crisis Relief program
to help government agencies and global NGOs run critical public service health
announcements during COVID-19. It also provided funding data on collaborative
initiatives such as the Alannah and Madeline Foundation’s Media Literacy Lab and
the number of students currently enrolled.
> Facebook provided expenditure data on a global funding round for academic
research into misinformation and polarisation. Two winners were located at
Australian universities. Facebook also provided expenditure data on grants
provided to their fact-checking partners to improve capacity during COVID-19.
Need for further data and metrics

The information signatories have provided is heavily focused on platform outputs –
measures that signatories have put in place – and on volumetric data relating to the
implementation of those measures. Where signatories provided comparative or
success-oriented data, this was often piecemeal or not directly related to actions under
the code:
> Google noted that changes to YouTube recommendation systems resulted in a
70% reduction in time watching non-subscribed, recommended content in the US
in 2019. They also noted that their aim is to have views of non-subscribed,
recommended borderline content (that is, content close to breaching content
policies) below 0.5%. However, no relation was drawn between these figures.
> Google provided data on YouTube video takedowns that showed the percentage
of videos removed before they were viewed and, separately, the percentage
removed for spam, scams or misleading content. While this may assist in
establishing a baseline, it does not cross-reference these 2 data points.
> Twitter’s report provided raw data on account actions, suspensions and deletions
relating to its misleading information policy. These will assist in establishing a
baseline for future reports but do not establish comparative metrics.
On the whole, reporting lacked systematic data, metrics or key performance indicators
(KPIs) that establish a baseline and enable the tracking of platform and industry
performance against code outcomes over time. Similar criticisms have been made
about the EU Code, and the European Commission has recommended the
development of both platform-specific and industry-wide KPIs. The ACMA has also
developed some high-level guidance at Appendix F to assist DIGI and signatories
ahead of their next annual report.
acma | 73
Australia-specific data
Signatories provided some data on the Australian context, but this was again
piecemeal, or not directly related to actions under the code.
> Twitter provided information on account suspensions and deletions for Australian
accounts but only global data on content labelling. Twitter noted that it has not
generally identified specific, large-scale, targeted information operations
originating outside of Australia and targeting people and conversations within
Australia.
> TikTok provided Australia-specific data on content labelling and removals and
views of the COVID-19 information hub (relevant to outcome 1a) but did not
provide data against other outcomes.
> Facebook provided data on the number of pieces of content removed from
Australian-specific pages and accounts between March and December 2020.
> Google provided data on the number of ads they blocked from Australian-based
advertisers for violating their misrepresentation ads policy. This includes ads
violating misleading representation, clickbait, and unacceptable business practices
policies. Google also provided raw figures on ads blocked or removed from
Australia-based advertisers relating to COVID-19. This included ads removed for
misleading claims, but also non-code-related matters such as price-gouging.
> Google provided data on the number of accounts removed for engaging in
coordinated influence operations. These are global removals and no information
was provided on the extent to which such operations affect Australian users.
While disinformation and misinformation are global issues, and signatories operate on
an international scale, reporting should include Australia-specific data and signatories
should establish a reporting regime against the Australian code.
The ACMA acknowledges that some signatories have provided confidential data
points to inform this report. While this has assisted in the development of this report,
we consider this data should have been included in published reports.
Trends in data
The reporting template asks signatories to provide qualitative or quantitative data on
trends where available. As noted above, there were some examples of case studies
where signatories provided empirical data as backing for the success of their
measures. For example, Redbubble reported data on the numbers of sales made for
content tagged by uploaders with anti-vaccination tags since April 2015.
However, there was generally a lack of long-term data in signatory reports, particularly
longitudinal quantitative data about the prevalence and types of disinformation and
misinformation or other key metrics, such as user behaviour.
Trend-related data would contribute to a greater understanding of the extent and

impact of disinformation and misinformation in Australia. Importantly, systematic
treatment of such data would create greater transparency in the effectiveness of
signatories’ actions in addressing emerging issues.
74 | acma
Finding 47: Reports provide some data on the Australian context, but this is often
piecemeal or not directly related to actions under the code. Reporting
should include Australia-specific data and signatories should establish a
reporting regime against the Australian code.
Finding 48: Reporting lacks trend-related data. Trend-related data would contribute
to a greater understanding of the extent and impact of disinformation
and misinformation in Australia.
acma | 75
6. Considerations for future
reform
Over the last 18 months, signatories have demonstrated their commitment to
addressing disinformation and misinformation by implementing a wide range of
measures and collaborating on the development and introduction of the code.
However, while the code enjoys broad support across industry, it is too early to draw
any concrete conclusions on its overall impact or effectiveness.
The building blocks for an effective self-regulatory scheme – such as complaints-

handling processes and other code administration matters – are still under
development. Broader public confidence is unlikely until the industry can demonstrate
it has built the necessary frameworks to effectively self-regulate.
As the code is still in its infancy, and due to be reviewed by signatories in early 2022,
the government could wait and see how the industry responds over the next 12
months. However, it may also wish to consider taking steps to strengthen regulatory
oversight and proactively address some of the issues identified in this report to
increase the overall likelihood of the code’s success.
In this chapter, we put forward recommendations to assist government in considering

its approach.
6.1. Consideration of findings

This report has made a range of findings on the code development process, the code
content and code framework. These include:
> The need for greater publicity when consulting on changes to the code, including
promotion through public communications channels and engagement with the
media.
> The suggested move to an opt-out code framework for code outcomes.
Signatories could provide statements against outcomes that do not apply to their
business model or services, rather than opting-in to each outcome individually.
> The scope of the code is limited by the narrow definition of harm and the
exclusion of some relevant products and services. This presents a risk that the
code will fail to provide adequate safeguards against the full range of potential
harms caused by the propagation of disinformation and misinformation on digital
platforms.
> The code reporting and administration framework is high level and yet to be
developed. Additional guidance outside of the code is yet to be developed, which
will be vital to the effectiveness of the code.
DIGI and signatories are encouraged to consider these findings when finalising their
code administration framework and reviewing the code in February 2022.

76 | acma
6.2. Continued oversight
The ACMA’s position paper provided a useful framework to guide the code
development process and the draft code. In releasing the draft code for consultation,
DIGI stated it had regard to this paper in the development of the code. Several
submitters also drew upon the paper in their submissions to the draft code –
suggesting elements of the ACMA’s guidance (such as expanding the code to cover
misinformation) should be included in the final draft. Subsequent discussions with
submitters indicated that the paper provided a good framework for distilling the issues
in an Australian context and was useful to inform the development of their
submissions.
As discussed in this report, the delay in finalising the code had follow-on effects on the
finalisation of the code administration framework and reporting guideline. Both the
framework and the guideline are still under development and are scheduled to be in
place by the end of August 2021. Recent discussions with DIGI indicate that
governance design is focused on incentivising signatories to meet the commitments
under the code and drive improvements over time. An independent committee with
defined terms of reference will be established to make sure signatories are meeting
their code commitments and will also have a role in the independent code review. DIGI
has advised that a key component of the model will be on arrangements to support
transparency reporting.
While DIGI has kept the ACMA updated on its progress, the information available is
extremely preliminary. This reflects the need to develop a completely new function by
an industry with limited experience in codes. More time and concrete information is
required for a considered assessment of code administration arrangements.
Given the status of code oversight arrangements, there is an argument that continued
oversight of the code is warranted. A continued oversight role would also allow the
ACMA to provide a more robust and fully formed assessment of the effectiveness of
the code once it has been in operation for a reasonable period of time. This is also
consistent with the government’s response to the DPI, which suggested the ACMA’s
June 2021 report would be the first of a number.
and should report to government on its effectiveness no later than the end of the 2022-
23 financial year. The ACMA should also continue to undertake relevant research to
inform government on the state of disinformation and misinformation in Australia.
It is possible that the ACMA’s findings will be addressed in the February 2022 code
review. However, many of the code deficiencies identified in this report were raised
with DIGI and the signatories throughout the code development process. DIGI’s
experiences in developing the code also suggest that it may be difficult to achieve
consensus on a range of code issues amongst current signatories.
The ACMA considers there is a continued role for guidance to publicly articulate clear
expectations where appropriate. Given the influential role that the ACMA’s position
paper had in the development of the draft code, this role should be continued.
The issuing of guidance has also been used in other jurisdictions to influence the
development of voluntary codes. For example, the European Commission recently
acma | 77
issued Guidance on Strengthening the Code of Practice on Disinformation to articulate
its expectations to platforms on the proposed revision of the EU Code. 113
Strengthening the EU Code
The European Commission issued its guidance in May 2021, following its
formal assessment of the EU Code in September 2020. 114 The guidance sets
out how platforms should step up their measures to address the shortcomings
identified in the assessment of the EU Code.
Some of the key recommendations from the guidance include:
> an expansion of the code to include misinformation in some areas, as

well as private messaging services
> wider participation from both established and emerging platforms, as well
as fact-checkers, content assessment organisations, and technology
developers
> wider participation from both established and emerging platforms, as well
as fact-checkers, content assessment organisations, and technology
developers
> a requirement for signatories to publicly justify their reason for opting out
of certain code provisions
> enhanced code provisions on the demonetisation of disinformation on

advertising channels.
It also suggests an enhanced monitoring and reporting framework which

includes the development of:
> platform-specific and industry-wide KPIs
> a harmonised reporting template to allow for cross-platform comparisons
> publicly accessible transparency centres created by platforms
> a new data access framework for the research community

> a dedicated taskforce consisting of representatives from signatories,
European Digital Media Observatory (EDMO), European Regulators
Group for Audiovisual Media Services (ERGA) and other relevant
experts, to evolve and adapt the code to technological, market, and
legislative developments.
The Commission will continue to oversee the strengthening the EU Code, with
a first draft anticipated in the European autumn 2021.
The continued oversight role will also allow the ACMA to assist signatories in the
development of more robust reporting arrangements. Improved reporting
arrangements also need to be supported by research to monitor the impacts of the
problem and better understand the effectiveness of platforms’ measures. The ACMA
has a strong existing research capability to understand this work, drawing upon our
existing work in the media and communications sector. This report could provide a
baseline for further research to monitor and understand this dynamic problem. The
continuation of targeted and relevant research would also be valuable to inform
114
European Commission, Commission Staff Working Document: Assessment of the Code of Practice on
Disinformation – Achievements and areas for further improvement, September 2020.
78 | acma
government of the state of misinformation and disinformation in Australia, and inform
future digital literacy initiatives.
6.3. Improving signatory reporting

As noted in Chapter 5, the initial reporting by signatories was inconsistent and, in
general, lacked the level of detail necessary to benchmark performance or assess the
effectiveness of individual platform measures. Without the identification of KPIs
supported by robust data, it will remain difficult for signatories to verify their progress
towards agreed code outcomes, and for industry to demonstrate the overall
effectiveness of the code.
The EC’s recently issued Guidance on Strengthening the Code of Practice on

Disinformation states the need for an enhanced reporting framework that includes
many of the same elements we have identified as lacking in the initial platform reports
under the code. 115 The EC also calls for specific improvements in general platform
transparency, including an obligation on platforms to create online transparency
centres that provide data against common industry KPIs.
The ACMA acknowledges that signatories’ initial reports represent their first attempt at
reporting under the code, and that there are several initiatives underway that could
improve reporting over time. Signatories are currently working towards agreement on a
consistent annual reporting format and approach, and there will be further
opportunities to review the strength of reporting processes during the upcoming 12-
month code review. The ACMA encourages platforms to consider establishing more-
formal collaboration processes to inform these reviews. The ACMA would welcome the
opportunity to provide additional guidance on these issues as the reporting guideline is
developed. Some initial guidance on reporting and measurement issues has been
provided at Appendix F.
However, the experiences of the EU code suggest this is a challenging issue, and that
signatories are likely to need further incentives to increase transparency. As
multinational corporations, internal policies and priorities may limit the extent to which
platforms are willing to invest in the necessary systems and resources to allow for
more detailed, Australia-specific reporting. Some signatories have indicated to the
ACMA that stronger regulatory backing would give them certainty and justify the
allocation of resources to generate reporting capability in new areas.

misinformation.
Formal information-gathering powers would allow the ACMA to better monitor the
progress of the voluntary code and incentivise behavioural change across industry
through greater transparency.
In France, the national media regulator Conseil supérieur de l'audiovisuel (CSA), has
formal information-gathering powers to oversee digital platforms, including the ability to
request specific data points and issue recommendations to platforms to improve
reporting. Platforms are also required to submit annual performance reports to CSA
115
European Commission, Guidance on Strengthening the Code of Practice on Disinformation,
26 May 2021.
acma | 79
outlining what measures they have taken to address disinformation in France. This
complements the voluntary code at the European level.
The ACMA’s consultations with the CSA suggest that this approach has improved
transparency of platform activities. Providing the ACMA with similar powers would
incentivise signatories to voluntarily make improvements to their annual reporting
against the code and allow the ACMA to compel individual digital platforms to provide
more detailed or robust data on their performance if their annual reporting is
insufficient. This information would also help inform future ACMA reports to
government on the ongoing effectiveness of the code and identify systemic industry-
wide issues that could inform targeted research.
The power to issue requests for specific information would be consistent with the
treatment of other industries that the ACMA regulates across the communications
sector, such as the ability to obtain information and documents from
telecommunications carriers under section 521 of the Telecommunications Act 1997,
or to obtain documents through investigations powers under section 173 of the
Broadcasting Services Act 1992.
Additionally, government could provide the ACMA with the ability to make record-
keeping rules for the digital platform industry, such as under section 529 of the
Telecommunications Act. Targeted record-keeping rules could require all digital
platforms to keep and regularly provide the ACMA with consistent and locally relevant
data that could be relied upon to monitor and analyse industry-wide changes and
developments over time. If this approach were taken, the ACMA would seek public
consultation on any proposed record-keeping rules.
6.4. Code administration

Code signatories and DIGI have demonstrated a great deal of commitment in the
development of a single, industry-wide code that has attracted strong industry support.
However, with the code administration framework not yet fully developed, areas of
concern remain.
Some stakeholders have raised questions about DIGI’s independence and lack of
transparency over its governance arrangements and funding. We note DIGI does not
publish any details about addressing conflicts of interests or the process for
membership. Transparency in these areas should be increased.
Major platforms are orders of magnitude larger than other regulated entities in the
sector and should therefore be prepared to adequately resource voluntary
initiatives. 116
The ACMA considers that the industry should be given additional time to bed down its
voluntary code. However, the risk remains that the current self-regulatory approach
may prove insufficient to incentivise broader behavioural change across industry, as:
> compliance with the current Code is uncertain at this stage given the data
provided by platforms
> it is not certain that current deficiencies with the code will be addressed by the
industry in its 12-month review
116
Four of the 8 signatories (Apple, Microsoft, Alphabet (Google’s parent company) and Facebook) are
within the top 10 most valuable publicly listed companies in the world by market capitalisation, each valued
at over $1 trillion USD. By comparison, Australia’s largest telecommunications provider, Telstra, ranks #610
globally; CompaniesMarketCap, Largest Companies by Market Cap, 29 June 2021.
80 | acma
> there are a range of non-signatories to the Code (see next section)
> usage of platforms may expand rapidly and new services introduced without these
being brought quickly into the Code’s remit.
In response, the government may wish to establish a ‘fall-back’ regulatory framework

to enable future intervention, if required to address non-compliance of new issues as
they emerge.
The ACMA currently has no regulatory powers to underpin its oversight role. The
establishment of a suite of reserve regulatory powers would allow the ACMA to take
further action if required. Actions could range from buttressing the current voluntary
Code with a registration process to incentivise industry to develop and enforce
compliance with codes, through to standards-making powers if a code fails to address
those harms. Such reserve powers would be defined by the government and may be
confined to issues of most concern.
Developing this reserve power framework would also improve incentives for industry to
improve voluntary arrangements and provide an appropriate backstop for further
action if required. Care would be needed to ensure the approach is responsive to the
risks of disinformation and misinformation, and the concerns of stakeholders. This can
be balanced by a robust public consultation process.
Non-signatories to the code

There is a risk that disinformation and misinformation on non-signatory services may
become a larger issue in the future. As evidenced in Chapter 2, some platforms have
experienced significant growth in Australia over the past 18 months. With major
platforms implementing measures to address misinformation and disinformation on
their services, purveyors of misinformation are moving to alternative platforms, like
Telegram, that are not code signatories and have less stringent content moderation
policies.
As the code is voluntary, there is currently no mechanism to compel platforms to sign

up to the code. In the first instance, consideration could be given to tailoring the code
to provide greater proportionality in reporting and other administrative matters for
smaller platforms. This may assist in encouraging these platforms to sign up
voluntarily.
More-formal options could be considered for platforms that do not participate in

voluntary arrangements or reject the emerging consensus on the need to address
disinformation and misinformation. If these platforms continue to grow their user-base,
they may present a higher risk to the Australian community.
This is an issue for both industry and government to consider and continued formal
monitoring and reporting is therefore recommended.
To address the risk of emerging platforms, the proposed EU Digital Services Act uses
a co-regulatory backstop that would enable the EC to enforce the participation of
particular platforms in a revised EU Code.
acma | 81
Providing the ACMA with reserve code-registration and standard-making powers such
as those outlined in Recommendation 4 would provide the government with the option
to act quickly to address potential harms on emerging platforms if required.
6.5. Improved collaboration mechanisms

A consistent theme in discussions about online disinformation and misinformation is
the complex, dynamic and multi-sided nature of the problem.
While platforms bear considerable responsibility for the quality of the information
environment on their services, a one-sided regulatory response that places sole
responsibility for addressing the problem on platforms is unlikely to be effective.
At a minimum, ongoing monitoring of online disinformation and misinformation is

required to provide oversight of a rapidly changing environment and a dynamic
industry where new and significant risks continue to emerge. Work is underway within
the Department of Home Affairs and the Department of Foreign Affairs and Trade to
fulfil this monitoring role on behalf of government.
However, there is also increasing recognition that formal frameworks for widespread
collaboration – that include, but extend well beyond, digital platforms – are critical to
addressing the problem. Collaborative frameworks can encourage greater information-
sharing between government, industry, academia and civil society to promote
increased understanding of the nature and extent of the problem and enable robust
assessment of the effectiveness of measures. They can also provide stronger
capability to monitor emerging risks and develop consensus-based approaches to
addressing them. Several submitters to the consultation on the draft code called for
greater government coordination in this area.
In the EU, the European Digital Media Observatory (EDMO) was established to fulfil
both a monitoring and coordination role. By providing support to independent
researchers and fact-checkers across 8 national hubs, EDMO will increase the
capacity to detect and analyse disinformation campaigns across member states. The
EC has called for EU Code signatories to establish formal cooperative mechanisms
with EDMO.

misinformation.
82 | acma
Given the range of issues identified above, the government may consider the value of
establishing a Misinformation and Disinformation Action Group to support
collaboration, cooperation and information-sharing on issues relating to disinformation
and misinformation. The Group could include digital platforms, government agencies,
interested researchers and NGOs. Key areas of focus for the group could include
identifying and monitoring emerging risks and data sharing to improve understanding
and inform awareness initiatives. 117 In particular, cooperation needs to be focused on
harms to users and on improving the overall online information environment. This is
consistent with approaches taken internationally and coordination with international
bodies like EDMO to develop consistent approaches would also be of great benefit.
6.6. Related areas of concern

There are several related areas of concern that government may wish to take into
account when considering any further responses to the code.
Misleading financial advertising

The dynamic nature of misinformation and disinformation and the flexibility of the code
contribute to a lack of clarity about what the code covers or should cover. For
example, recent media reporting suggested that the code would cover misleading
financial advertising in the context of the Mayfair 101 investment scheme. 118
Signatory reports from Google and Microsoft indicated their advertising policies would
capture this type of advertising. However, these policies did not prevent advertising by
the Mayfair 101 investment scheme.
The ACMA notes also that the code is intended to address harms arising in specific
areas of platform responsibility. As a voluntary, outcomes-based code, it places no
obligations on platforms to take particular actions. In addition, the misleading financial
advertising in the Mayfair 101 case may not reach the threshold of having a
reasonable likelihood of serious and imminent harm to be considered misinformation
under the code.
The code may, therefore, not be the appropriate vehicle for establishing protections in
areas that are not directly related to the online environment, where principles such as
the freedom of expression are not a primary concern. The government may wish to
consider whether this issue should be considered in the context of changes to the
Australian Securities and Investment Commission Act 2001, Australian Consumer Law
or other associated regulation.
Micro-targeted advertising
The 12-month review should consider extending the code to explicitly cover related
issues of platform responsibility such as misleading advertising using micro-targeting.
A similar extension has also been contemplated in the EC’s Guidance on
Strengthening the Code of Practice on Disinformation.
The ACMA intends to continue monitoring this area to identify and assess any areas of
harm that may not be addressed by the code or related initiatives.
117
Montgomery, Disinformation as a Wicked Problem: Why We Need Co-Regulatory Frameworks,
Brookings Institute, 2020, offers a detailed discussion of how such mechanisms might be approached.
118
Davidson, J., ‘Review of tech disinformation code promised after Mayfair 101 scandal’, Australian
Financial Review, 12 April 2021.
acma | 83
News quality initiatives
Some news quality initiatives (such as the obligation to develop an original news
proposal) are included as part of the minimum standards of the news media bargaining
code. These initiatives do not become an obligation on a platform until it is designated
under the news media bargaining code. Some stakeholders have indicated concern
that news quality-related initiatives may fall between the 2 codes. While there is a
range of activities underway targeted at improving news quality (see Appendix E),
consideration could be given to addressing any gaps via the provision of formal
guidance during the review of the disinformation and misinformation code.
Extension of code beyond digital platforms

Consideration could also be given to extending the scope of the code beyond digital
platforms. The EU Code includes advertising bodies as signatories, and the recent EC
guidance recommends broader participation from the advertising industry to increase
the code’s power to drive the de-monetisation of disinformation. It could also be
extended to include local fact-checking organisations, content assessment
organisations, and those providing tools and solutions for fighting misinformation. We
note that Adobe has already signed up to the code, and other technology providers
could follow Adobe’s lead.
84 | acma
Appendix A: Full list of
recommendations and findings
List of recommendations
and should report to government on its effectiveness no later than the end of the
2022–23 financial year. The ACMA should also continue to undertake relevant
research to inform government on the state of disinformation and misinformation in
Australia.

misinformation.

misinformation.
List of findings
Finding 1: Most Australians are concerned about, and have experienced, online
misinformation. Higher exposure is associated with heavy use of digital
platforms, disproportionately impacting younger Australians.
Finding 2: Access to authoritative and trusted sources of news and information is
an important mitigation against misinformation. Those that rely on social
media as a main source of news have a greater likelihood of being
misinformed about COVID-19.
Finding 3: Given its nature and the ongoing challenges in accessing relevant data,
the true scale and volume of misinformation in Australia is currently
unknown.
Finding 4: Australians report seeing the most amount of misinformation on large
platforms such as Facebook and Twitter. However, private messaging
services and smaller platforms with less strict content moderation
policies, like Telegram, are also being embraced by conspiracy-oriented
communities.
Finding 5: Misinformation typically stems from small online conspiratorial
communities, but can be amplified by influential individuals, digital
platform design, as well as the media.
acma | 85
Finding 6: Conspiratorial content is designed to be highly engaging, fuelling
outrage, and building on a sense of community. The confluence of
conspiracy theories around COVID-19 has created more paths to online
misinformation.
Finding 7: There is some evidence of co-ordinated inauthentic activity surrounding
popular misinformation narratives in Australia. Those who spread
misinformation often seek to reframe global conspiratorial narratives,
like QAnon, in a local context.
Finding 8: Misinformation narratives can result in a wide range of acute and
chronic harms, including the erosion of trust in authoritative sources and
democratic institutions over time.
Finding 9: The real-world consequences of misinformation have been readily

apparent over the past 18 months: inciting violence, undermining official
health advice, and causing tangible financial impacts on governments,
industry and consumers.
Finding 10: Most Australians are aware of platform measures to remove or label
offending content, but few have direct experience. Early evidence
suggests these steps have been somewhat effective in reducing
amplification of misinformation on particular platforms.
Finding 11: Australians see the issue of misinformation to be one of joint
responsibility – split between individual users, platforms, and
government. There is some scepticism in the ability of platforms to self-
regulate, and concern about government’s role in regulating speech.
Finding 12: Information on the effectiveness of platform measures is limited, and
more needs to be done to better understand what measures work and
to monitor the effectiveness of platform moderation activities.
Finding 13: In leading code development, DIGI successfully managed a novel,
complex and time-sensitive project, navigating a range of competing
interests across a disparate group of stakeholders that included both
members and non-members of DIGI.
Finding 14: DIGI undertook a meaningful public consultation process on its draft
code, generating a range of feedback from academia, industry, and
parts of civil society, which visibly informed the final code.
Finding 15: DIGI could have improved its consultation process with greater publicity,
including promoting it through existing public communications channels
and engagement with the media.
Finding 16: DIGI dealt with stakeholder feedback in a relatively open and
transparent manner. However, the significant change in scope meant it
would have been best practice to provide stakeholders a further
opportunity to comment on the final drafting prior to finalisation.
Finding 17: The bulk of ‘major platforms’ in Australia have signed up to the code. As
such, it should be regarded as an industry-wide initiative.
Finding 18: DIGI should continue to encourage other popular platforms, like
Snapchat and Reddit, to sign up to the code, even if they do not meet
the proposed threshold of one million active monthly users. DIGI should
actively publicise the involvement of any additional code signatories as
soon as practicable after their signing.
Finding 19: Industry participants should consider the role of private messaging
platforms and smaller alternative platforms in the amplification of
disinformation and misinformation and explore options for how these
platforms could be included within the code framework.
86 | acma
Finding 20: DIGI has developed an outcomes-based code that has allowed
platforms with a range of business models to sign up to a single code.
Finding 21: The code objectives and principles meet the government objective of
striking a balance between encouraging platform interventions and
protecting freedom of expression, privacy and other rights
Finding 22: The code should be strengthened by taking an opt-out approach. Opting
out of an outcome should be permitted only where the outcome is not
relevant to the signatory’s services. Signatories should provide
adequate justification when opting out.
Finding 23: The code covers both disinformation and misinformation. This is one of
the key strengths of the code, and is an improvement on the current
EU Code.
Finding 24: The definition of harm in the code is too narrow to provide adequate
safeguards against the full range of harms caused by the propagation of
disinformation and misinformation.
Finding 25: Private messaging services should be included within the scope of the
code as these are known vectors of disinformation and misinformation.
These should be included with appropriate caveats on the right to
privacy.
Finding 26: The code should clarify that the exclusion of professional news content
applies only to the application of counter-misinformation measures. It
should also clarify that news aggregation services are in scope.
Finding 27: The treatment of paid and sponsored content should be clearer in the
code. This should include a clear definition of sponsored content.
Finding 28: In addition to improving public awareness of the source of political

advertising, the code should also cover the source of issues-based
advertising.
Finding 29: The output-focused framing of several objectives and measures may
encourage reporting to focus on outputs rather than progress towards
code outcomes.
Finding 30: The code should include industry-wide frameworks for the development
and implementation of individual platform measures. Examples could
include frameworks to establish:
> criteria for the assessment of harm
> criteria for assessing news and information quality
> processes for the exchange of information between platforms on
disinformation and misinformation risks
> commitments to address the propagation of disinformation and
misinformation via platform advertising channels
> commitments to address the risks of propagation via platform
algorithms and architecture.
Finding 31: The code provides a high-level code administration framework. Given
that detailed arrangements for code administration, compliance with the
code, and consumer complaints are still under development, the
ACMA’s ability to assess their practical effectiveness is constrained.
acma | 87
Finding 32: The code should include a framework setting out principles for the
structure and operation of the sub-committee to provide greater
transparency and accountability.
Finding 33: The reporting template provides a workable foundation for the reporting
guideline. Reporting should incorporate adequate data to measures
performance against KPIs under each outcome; detailed action plans to
address areas identified for improvement; and a clearer distinction
between measures (that is, outputs) and the effectiveness of these
measures (progress towards outcomes).
Finding 34: The proposed 12-month review will provide an opportunity for
findings from this report, and other developments, to be incorporated
into the code.
Finding 35: The lack of detail on code administration matters, including on the
operation of the sub-committee and guidelines for future code reporting,
has limited the ACMA’s ability to undertake a full assessment on the
likely effectiveness of the code.
Finding 36: On the whole, signatories have met the initial reporting requirements set
out in the code.
Finding 37: For the most part, signatories have provided appropriate explanations
where they have not opted-in to specific commitments.
Finding 38: A more uniform approach to reporting would assist in cross-platform

assessment and increase transparency of platform measures and
performance.
Finding 39: For future reports, signatories should clearly specify the products and
services covered by the code, and justify any major exclusions.
Finding 40: Signatories have a wide range of measures in place to address the
problems of disinformation and misinformation and to improve the
quality of news and information on their services. They also
demonstrate responsiveness to significant changes over the last 18
months, as well as to public and government calls for stronger action.
Finding 41: It is expected that signatories will develop more Australia-focused

measures over time.
Finding 42: In general, the reports are heavily focused on current measures and
past actions, and signatories have provided little systematic information
on future initiatives. In some cases, it is not clear to what extent certain
measures will contribute to the achievement of the code outcomes
under which they have been reported.
Finding 43: There are inconsistencies in the interpretations of key terms between
signatories, which are drawn from pre-existing definitions from their
internal, often global, policies. This makes it difficult to interpret and
assess performance and to make industry-wide comparisons.
Finding 44: A harmonised template would assist in comparing initiatives across

platforms. It would also allow clear reporting on additional information
beyond the requirements of the code.
88 | acma
Finding 45: Signatories have provided a large range of information on the actions
they have taken to address misinformation, disinformation and news
quality and to invest in collaborative initiatives. This demonstrates
signatories’ commitment to addressing these issues.
Finding 46: The information signatories have provided is heavily focused on

platform outputs and on volumetric data. Reporting lacks systematic
data, metrics or key performance indicators (KPIs) that establish a
baseline and enable the tracking of platform and industry performance
against code outcomes over time.
Finding 47: Reports provide some data on the Australian context, but this is often
piecemeal or not directly related to actions under the code. Reporting
should include Australia-specific data and signatories should establish a
reporting regime against the Australian code.
Finding 48: Reporting lacks trend-related data. Trend-related data would contribute
to a greater understanding of the extent and impact of disinformation
and misinformation in Australia.
acma | 89
Appendix B: Signatory
assessment reports
Australian user base Unknown

Services covered: All Adobe services able to take
advantage of the Content Authenticity
Initiative (CAI), including Creative Cloud
Code commitments: 5.8, 5.9, 5.18, 5.19, 5.28, 5.29, 5.30
Strengths
> Adobe has signed up to the code based on the CAI, opting into the outcomes
that it considers proportionate and relevant to its business.
> The report outlines the importance of the CAI and associated C2PA standards to
improve attribution of digital content, both for users of Adobe products, and more
broadly across industry. The level of information provided in the report is
appropriate given Adobe’s limited ranges of commitments under the code.
> The inclusion of software providers, like Adobe, strengthens the code by
broadening its remit to include those who can provide tools and solutions for
addressing disinformation and misinformation. This may encourage greater
collaboration on common solutions between platforms and providers.
Weaknesses
> No Australia-specific measures are identified in the code, beyond briefings and
events to promote consideration of content authenticity and increase
understanding of attribution. Similarly, Adobe provides no Australia-specific data.
> While Adobe has a goal for the future of the CAI, it doesn’t provide any concrete
data, discuss targets, or identify what metrics it will use to measure success.
> Adobe provides an overarching rationale for its limited involvement, but it
remains unclear why it has chosen not to opt-in to some of the code outcomes
that appear relevant to the CAI, such as developing partnerships and
collaborating on strategic research (Outcome 6).
Recommendations
Adobe should consider publishing information on its measures of success under the
code, such as targets for the uptake of CAI in Australia. Adobe should also consider
whether it could opt-in to other code outcomes that appear relevant to the CAI.
90 | acma
Apple

Services covered: Apple News
Code commitments: 1a, 1d, 2, 4, 6, 7 (5.28 only)
General comment
Apple’s commitments under the code are limited to its Apple News service. While
we welcome Apple’s involvement, certain aspects of its commitment remain
unclear, due to the code’s treatment of news aggregation services and the
exclusion of professional news content from the operation of the code.
Strengths
> Apple has opted-in to several commitments in relation to Apple News. For most
of the provisions it has opted-out of, it has provided a short explanation for why
these provisions are not applicable to Apple News.
> Apple’s report is detailed and contains several Australian-specific examples and
commitments. These include its approach to the 2019-20 Australian bushfire
season and the employment of Australian journalists and editors at Apple News.
Weaknesses
> It is unclear whether some measures in the report are existing or future
initiatives.
> The report provides minimal data on Apple’s content moderation activities. The
one data point on content removal is not limited to disinformation and
misinformation, and it is a global statistic.
> Apple did not include a reason why it has chosen not to opt-in to all measures
under Outcome 7, including publishing additional information detailing its
progress against implementing code commitments.
Recommendations
Apple should consider providing more Australian-specific data to demonstrate how
its relevant policies and initiatives are helping address the issues of disinformation
and misinformation.
acma | 91
Australian user base: 17 million monthly active users
(Facebook), 9 million monthly active
users (Instagram)
Services covered: Facebook, Instagram
Code commitments: All code outcomes
Strengths
> Facebook sets out 43 specific commitments under the 7 code outcomes.
Commitments include 4 new Australian initiatives – 2 due to be rolled out in
2021. The expansion of its political-advertising policy to social-issue
advertisements commenced on 29 June 2021.
> The report provided detailed explanations of Facebook’s global policies and
initiatives for combatting misinformation and disinformation, including:
> coordinated inauthentic behaviour (disinformation) policies
> a graduated approach to harmful misinformation which has adjusted in
response to more recent developments
> measures to promote authoritative information and provide users with tools to
help them assess the quality of sources and factual content.
> It also includes Australia-specific data and information on its recent actions
relating to COVID-19 misinformation, and information on its investment in
research and collaborative initiatives including funding academic research,
supporting local events and developing misinformation-focused training.
> Facebook was the only signatory to broadly publicise the release of its report.
Weaknesses
> Reported data is piecemeal, and Facebook does not establish any metrics that
could be used to track the effectiveness of its measures, particularly in the
Australian context. It is noted that this will be developed over time.
> While Facebook provides some information on its WhatsApp and Messenger
measures, it does not consider these services to be covered under the code. It is
unclear whether Facebook will continue reporting on these services.
> The report adopts internal definitions, with disinformation being used to refer to
inauthentic behaviour with the intention to deceive, and misinformation to refer to
content that is false or misleading. This is inconsistent with the code.
Recommendations
Facebook should provide more-extensive data against robust KPIs that can be used
to assess the effectiveness of its measures. Although private messaging services
are currently excluded, Facebook could consider nominating its messaging services
under the code to provide greater transparency over its existing activities.
92 | acma
Australian user base 18 million unique monthly users (Google
Search), 16.5 million unique monthly
visitors (YouTube)
Services covered: Google Search, YouTube, Google Ads
Strengths
> Google signed up to all outcomes under the code and provided a detailed initial
report highlighting a range of relevant policies and measures.
> The report outlines the categories of intervention, its relevant global policies, and
the mechanisms for users to report violations against these policies.
> Google identifies several Australia-specific measures, such as new policies and
identity verification requirements for Australian election ads, and funding of
Australian research, including through the Google News Initiative.
> The report provided a detailed case study on steps taken to address COVID-19
misinformation, including promoting official health advice from the Australian
Government, providing ad grants to the government and local NGOs, and
partnering with local YouTube creators on public service announcements.
> Google noted that it views its commitments under the code as a floor rather than
a ceiling and will strive for continual improvement over time.
Weaknesses
> There is no identification or discussion of future measures, or what continual
improvement could look like in practice under the code.
> Google provides limited evidence demonstrating the effectiveness of its
existing measures, or no data on trends relating to disinformation or
misinformation over time.
> The reports don’t provide a rationale as to why some but not other Google
services are out of scope. The report includes some information on measures
relating to Google News but it isn’t clear if Google News is covered under
the code.
> Google published limited data directly related to Australia, despite comparable
breakdowns being publicly available for other countries (e.g., via YouTube
Community Guidelines enforcement).
Recommendations
For future reporting, Google should seek to provide further detail about how it
measures success, publicly report on more Australia-specific data, and discuss
future initiatives or planned improvements under the code.
acma | 93
Australian user base 6.5 million monthly active users
(LinkedIn), 1.7 million monthly active
users (Bing)
Services covered: All consumer-facing products
Code commitments: All except 5.28 (undecided)
Strengths
> Microsoft has committed to nearly all outcomes under the code, across all of its
consumer-facing products.
> Microsoft provides a broad list of relevant programs and initiatives that it has
undertaken to address the issues of disinformation and misinformation.
> The report includes some recent actions and data relating to COVID-19
advertising, and details investment into emerging issues, such as data voids
and deepfakes.
Weaknesses
> It is not clear which Microsoft services are covered by the code. Microsoft has
said its report covers its ‘consumer-facing services’ (referencing Bing, LinkedIn,
Microsoft News, and Microsoft Advertising), however the extent of this service-
type is nebulous. While the ACMA appreciates this could possibly extend to a
growing list of hundreds of services, a more clear and specific scope or list of
services covered would be beneficial.
> Data does not establish any metrics that could be used to track the effectiveness
of its measures. Of the minimal data provided, it is not always used to show how
effective a relevant policy or initiative is. The published report also does not
provide any Australia-specific data points.
> While Microsoft refers to measure 5.27 in its annual report, its commitment is left
blank in the nomination form. In discussions with the ACMA, Microsoft said it has
yet to decide whether the opt-in to 5.27 and is awaiting further detail from DIGI.
This could have been made clear in its nomination form.
> The report references Microsoft’s ‘On the Issues’ blog as a place where
company announcements about technology policy issues, including those
relating to disinformation and misinformation, are hosted. There is currently no
reference to the Australian code on this site.
Recommendations
To improve transparency, Microsoft should specify the services it considers to be
covered by the code. Microsoft should also consider providing more Australia-
specific data to build KPIs and demonstrate how its relevant policies and initiatives
are helping address the issues of disinformation and misinformation domestically.
94 | acma
Services covered: Redbubble
Code commitments: All except 1(d), 4, 5 and part of 2. As an
online marketplace, Redbubble has
opted out of Outcomes 4 and 5 as it
does not provide news, factual content
or third-party advertising.
Strengths
> The report is clear and consumer-focused, and Redbubble provides clear and
reasonable justifications for opting out of certain code outcomes.
> Redbubble provides trend-related data on misinformation on its platform,
including anti-vaccination content, and a general summary of future initiatives
relating to misinformation and disinformation.
> The report provides a clear explanation of relevant policies and how these are
implemented.
> Redbubble uses credible and trusted sources, including independent fact
checking sites, to determine the boundaries of disinformation and misinformation
and compile guidelines. Precedent and global context are also used to inform
decisions and policies.
Weaknesses
> Redbubble has not opted into Outcome 1(d) (provide general information on
actions taken in response to user reports), for the reason that reporting is
currently anonymous and no mechanism for publishing Redbubble’s responses
is currently available.
> Redbubble provides only a general commitment to research and collaboration
under Outcome 6 and no further commitment to transparency under Objective 7
beyond the code report. However, Redbubble has a small Australian user base
and limited exposure to disinformation and misinformation.
> Redbubble’s policies include a definition of harmful misinformation, but do not
refer to disinformation as defined in the DIGI code. However, Redbubble’s
definition of misinformation appears to be broad enough to cover both concepts
and their report includes information on its measures relating to demonetisation
and service integrity under Outcomes 2 and 3 respectively.
Recommendations
As Redbubble has a user-reporting function for reporting misinformation, it should
consider opting into Outcome 1(d) and working towards the provision of suitable
transparency mechanisms under the code. Redbubble could also consider
involvement in more formal research and collaboration partnerships to feed into its
content policy frameworks. To further strengthen its future reporting, Redbubble
could consider including data on the actions it has taken against disinformation and
misinformation content.
acma | 95
Services covered: TikTok
Strengths
> As a smaller platform and more recent entrant to the Australian market, it is
pleasing to see TikTok sign up to all code outcomes
> TikTok has provided some quantitative Australian-specific data on COVID-19
and medical misinformation video removals, and numbers of unsubstantiated
COVID-19 claims tagged with information notices relevant to Outcome 1a.
> The report includes month-by-month breakdown of data, allowing for a deeper
insight into the effectiveness of some of its policies over time.
> The report followed the agreed reporting template, providing sufficient contextual
information about its approach to disinformation and misinformation, before
reporting on specific against code outcomes.
Weaknesses
> TikTok’s report does not fully align with the definitions of disinformation and
misinformation used in the code. TikTok’s definition of misinformation, for
example, does not capture misleading information, except to the extent that it
relates to elections or civic processes and synthetic or manipulated media.
> TikTok does not clearly articulate its approach to assessing harm and deciding
whether content should be labelled or removed.
> It is not clear how TikTok’s reported measure against Outcome 6 (Asia-Pacific
Advisory Council) will contribute to the achievement of the code outcome.
Recommendations
TikTok should provide greater detail how its policies and broader measures align to
the code outcomes, and better signal what future initiatives it is considering,
particularly in relation to Australian users.
96 | acma
Twitter

Services covered: Twitter
Code commitments: All code outcomes, except Outcome 5
Strengths
> Twitter has opted into all code outcomes except Outcome 5 on the basis that it
does not accept political advertising.
> The report shows meaningful commitment to action against misinformation in the
context of specific issues such as COVID-19 and elections, and more general
commitment surrounding disinformation, information quality, and public
transparency. Twitter provides some detailed information on its policies and tools
on disinformation and misinformation under Outcome 1.
> Twitter provides some specific data on its actions against misinformation and
disinformation, including account suspensions, removals and content removals,
both globally and for Australian accounts.
> COVID-19 policy evidences a responsive, proportionate, risk-based approach.
Weaknesses
> The report does not address how the problems of disinformation and
misinformation manifest on the platform, with no data on trends or prevalence
that would facilitate benchmarking.
> While Twitter has shared some information with the ACMA on a confidential
basis, it would be beneficial for more information on future measures in response
to the code or broader environmental developments, either globally or in
Australia were included in the report.
> The data provided is not sufficiently comprehensive or specific to
Australia. For example, the report gives no data on actions taken against
synthetic and manipulated media and provides only global numbers on content
labelling for misinformation.
> Does not address how the principles of the COVID-19 misleading information
or civic integrity policies might be applied to other areas of potential harm,
or whether their user-reporting mechanism for manipulated media will be
extended to other areas of disinformation and misinformation in accordance
with Outcome 1c.
Recommendations
Twitter should strengthen its reporting to include KPIs that track performance under
the code, detailed Australia-specific data, trends, and planned measures. Twitter
should consider expanding the risk-based principles in its COVID-19 misinformation
policy to cover misinformation more generally and its user-reporting mechanism to
cover the scope of misinformation and disinformation set out in the code.
acma | 97
Appendix C: Timeline of key
events
Since the government published its response to the DPI in December 2019, digital
platforms have implemented a range of changes to address disinformation and
misinformation on their platforms. The following timeline traces these activities against
key social and political events that have impacted disinformation and misinformation
trends. Platform responses include policy changes, enforcement actions, changes to
the platform features and functions, and other actions and initiatives.
Almost all policy changes and new initiatives identified by the ACMA have been in
response to either the COVID-19 pandemic or the 2020 US presidential election.
In high-level summary, platforms have:

1. invested in third-party fact-checking organisations to proactively identify and flag
false information on their service
2. proactively updated their policies to specifically address unique events where
there is a heightened risk of harm and increased enforcement actions against
potentially misleading and false information in relation to unique events
3. invested in technological means to signal credible, relevant and authentic
information
4. signalled information that may be false or misleading and provided additional
context by linking to authoritative information sources, such as those published by
World Health Organisation (WHO)
5. provided financial assistance and grants to news outlets and government and not-
for-profit organisations to bolster the spread of credible information and news
6. increased detection, monitoring and enforcement action against groups and

networks who use their services to spread disinformation and misinformation.
Timeline key
Significant social or political events
Policy changes
Enforcement actions
Changes to the platform features and functions
Other actions or initiatives.
Australian events or platforms measures directly relevant to Australia
98 | acma
December 2019
12 Dec Australia – Australian Government responds to the ACCC’s Digital Platforms

Inquiry; calls on digital platforms to establish a voluntary disinformation code.
31 Dec China – First cases of a novel coronavirus reported in Wuhan.
January 2020
Early Australia – South-eastern bushfires (‘Black Summer’) at peak. Fires in NSW

Jan and Victoria are extinguished or contained by early March.
6 Jan Facebook announces changes to its Manipulated Media Policy to remove

videos that are the product of machine learning techniques (e.g., deepfakes)
and have been edited or synthesised in a manner likely mislead an average
person to believe a subject of the video said words they did not say. 119
8 Jan TikTok publishes expanded and more detailed Community Guidelines

intended to be easier to understand and enforce. 120 The guidelines prohibit
content intended to mislead, including synthetic content (e.g., shallow or deep-
fakes), spam and disinformation. 121 Misinformation that may cause harm,
including harm to an individual’s health, or that may mislead the public about
elections and civic processes, is also prohibited. 122
13 Jan Thailand – First cases of coronavirus outside China reported.
25 Jan Australia – First local case of coronavirus reported.
29 Jan Canada – Canada’s Communications Future: Time to Act report recommends

legislative reform to address harmful content on digital platforms.
Twitter launches a new search prompt, in partnership with authoritative health

agencies (including with the Australian Department of Health) to help surface
credible COVID-19 information and remove auto-suggest options likely to lead
to non-credible information. 123
30 Jan Global – WHO declares the COVID-19 outbreak a Public Health Emergency
of International Concern.
Late Facebook announces it will apply its existing Misinformation and Harm policy
Jan (covering Facebook and Instagram) to remove COVID-19 misinformation that
could cause physical harm. For example, claims that drinking bleach can cure
the virus, or that social distancing is ineffective. Facebook has had policies in
place to remove misinformation that can lead to serious and imminent physical
harm since January 2018. 124
119
Monika Bickert, Enforcing Against Manipulated Media - About Facebook.
120
TikTok submission to the Senate Select Committee on Foreign interference through social media.
121
TikTok, Community Guidelines.
122
TikTok, Building to support content, account, and platform integrity.
123
Jun Chu and Jennifer McDonald, Helping the world find credible information about novel #coronavirus.
124
Facebook’s Submission to the Senate Select Committee on Foreign Interference through Social Media,
p 9.
acma | 99
Facebook updates its policies to limit the spread of information that would
otherwise not contravene its Community Standards, but that promote
misleading information about the pandemic or discourages vaccination. 125 It
will remove false information that falls into a number of broad categories of
COVID-19 related misinformation: false information on the existence or
severity of COVID-19, links to 5G technologies, transmission and immunity,
cures or prevention methods and information that discourages good
health practice.
Google treats the COVID-19 pandemic as a ‘sensitive event’ for the purposes
of applying its policies, which variously prohibit behaviours and content-related
to disinformation and misinformation. 126
TikTok makes a number of specific updates to its policies in response to the

COVID-19 pandemic, more readily remove or limit COVID-19 misinformation,
including to: 127
> remove false medical advice that may cause harm
> remove false information that is likely to cause panic
> limit the dissemination of false claims that the pandemic was
deliberately spread
> remove content that suggests certain groups of people are more likely to
have or spread COVID-19.
TikTok places restrictions on advertising on its platform, including a

prohibition against advertising that advocates against vaccination. 128
Redbubble publishes guidance on coronavirus/COVID-19 content. The

guidance sets out certain types of COVID-19 related content that may be
removed from the site, including:
> messages which discriminate against certain groups
> jokes at the expense of victims
> works with ambiguous or harmful intent
> unnecessarily graphic content
> messages spreading false information or causing panic. 129
February 2020
Facebook conducts a ‘Strategic Network Disruption’ against the Proud

Boys. 130 This involves removing individual and group accounts to prevent
extremist groups from using Facebook as an organising tool. Facebook
repeats this exercise in June and September 2020.
Apple News launches special coverage of the 2020 US presidential election

for US users. It includes a ‘news literacy guide’ developed in partnership with
the News Literacy Project, to help users identify misinformation. 131
1 Feb UK – EU–UK Withdrawal Agreement comes into force.
125
Instagram, COVID-19 and Vaccine Policy Updates and Protections.
126
Google, Google Initial transparency report, May 2021.
127
TikTok, submission to the Senate Select Committee on Foreign Interference Through Social Media.
128
TikTok, Supporting our community through COVID-19.
129
Redbubble, Coronavirus/COVID-19 Content on Redbubble.
130
Facebook, Taking action to combat misinformation, polarization, and dangerous organizations.
131
Apple Newsroom, ‘Apple News launches special coverage of the 2020 presidential election’.
100 | acma
4 Feb Twitter publishes its policy on synthetic and manipulated media. 132 The policy
indicates this content is prohibited and subject to a range of enforcement
actions, such as labelling. 133 Synthetic or manipulated media that is likely to
cause harm, such as physical harm or civil unrest, will be removed. Users may
report content they believe may contravene this policy.
24 Feb Instagram announces new measures to support accurate COVID-19

information, including educational resources and limiting COVID-19 accounts
accessible in recommendations to credible health organisations. 134
March 2020
YouTube introduces COVID-19 information panels on its home page and on

pages related to the pandemic. 135
Apple is reported to only be approving apps related to COVID-19 if they are

developed by official institutions. 136 Apple reviews apps for release on its app
store under its App Store Review Guidelines. 137
TikTok opens a US-based Transparency Centre, which will enable

independent experts to observe how TikTok implements its content
moderation and complaints-handling policies. 138
3 Mar Facebook implements a new policy to prevent advertisers from exploiting the
pandemic for financial gain and introduces information pop-ups on top of
search results directing users to the WHO and local health authorities to assist
users to find reliable information on the COVID-19 pandemic.
5 Mar TikTok launches a COVID-19 Information Hub in the UK. Users who explore
hashtags related to COVID-19 pages are directed to the Hub, which includes
information from trusted sources, such as the WHO and the Red Cross. 139
11 Mar Global – WHO declares COVID-19 a pandemic.
13 Mar Australia – National Cabinet formed in response to the COVID-19 pandemic.
17 Mar Facebook, Google, LinkedIn, Microsoft, Twitter, YouTube and Reddit

issue a joint statement outlining their approach to addressing misinformation
about COVID-19 on their respective platforms.
17 Mar Facebook launches a US$1 million grant program with The International Fact-
Checking Network and a further US$1 million worth of grants to local news
132
Twitter, Synthetic and manipulated media policy.
133
Achuthan, A. and Roth, Y. Building rules in public: Our approach to synthetic & manipulated media.
134
Instagram, Keeping people safe, informed and supported on Instagram.
135
Google, How you'll find accurate and timely information on COVID-19 vaccines.
136
Kif Leswing, ‘Apple rejecting coronavirus apps that aren’t from health organisations, app makers say’,
CNBC.
137
Apple, App Store Review Guidelines.
138
Sarah Perez, ‘TikTok to open a ‘Transparency Centre’ where outside experts can examine its moderation
practices’, Techrunch.
139
TikTok submission to the Senate Select Committee on Foreign Interference Through Social Media.
acma | 101
organisations to support costs related to covering the COVID-19 pandemic. 140
If a piece of content is rated false by fact-checkers, Facebook will limit its
distribution and label it with a warning and information to provide more
context. Users are unable to view such content without clicking through the
warning label. Facebook states that in 95% of cases, users will not proceed to
view the false content. 141 Facebook states that: ‘in April 2020 alone, we
applied the label and reduced the distribution of more than 50 million posts
worldwide, based on more than 7,500 fact-checks’. 142
18 Mar Facebook launches the Facebook Coronavirus Information Center, providing

a centralised location for users to access news and information about COVID-
19. In Australia, the Center includes official Australian Government
information.
18 Mar Twitter broadens its the definition of harm to address content that contradicts
guidance on COVID-19 from ‘authoritative sources of global and local public
health information’. 143 Under Twitter’s COVID-19 misleading information
policy, content may be labelled or removed if it definitively advances a false
claim, is demonstrably misleading and likely to cause serious harm. 144 There
are 5 categories of information covered by the policy, include false or
misleading information about:
> the nature of the virus
> the efficacy and/or safety of preventative measures, treatments, or other
precautions to mitigate or treat the disease
> official regulations, restrictions, or exemptions pertaining to health
advisories
> the prevalence of the virus, or risk of infection or death.
Twitter provides local authorities and experts with access to a Partner Support
Portal to assess information and escalate problematic content for human
review. 145 Content may be removed under this policy and repeated
contraventions may result in an account being permanently suspended. The
policy also makes exemptions for commentary, satire, counter speech,
first-hand accounts or public debate for the purposes of advancing COVID-19
science and research.
20 Mar Australia – National borders closed to non-residents.
20 Mar US – President Trump claims that hydroxychloroquine could prevent or treat

COVID-19.
20 Mar Facebook launches the WHO Health Alert on WhatsApp. The alert
provides reliable information on matters such as how to prevent the spread of
COVID-19 and correcting false information and conspiracy theories. The alert
140
Facebook, Keeping People Safe and Informed About the Coronavirus.
141
Facebook, submission to the Senate Select Committee on Foreign Interference through Social Media,
p 10.
142
Facebook, Facebook response to the Australian disinformation and misinformation industry code, p. 19.
143
@Vijaya and Matt Derella, An update on our continuity strategy during COVID-19.
144
Twitter, COVID-19 misleading information policy.
145
Twitter, Australian code of practice on disinformation ad misinformation initial report, p.14.
102 | acma
is first made available in English, and then rolled out in Arabic, Chinese,
French, Russian and Spanish. 146
24 Mar Facebook announces a range of updates to improve the accuracy of

COVID-19 information on Instagram, including:
> educational resources
> labels on accurate authoritative information
> only recommending COVID-19 accounts from credible health
organisations. 147
29 Mar Australia – COVID-19 social distancing rules strengthened, establishing

2-person limit for all gatherings, and new lockdown restrictions.
30 Mar UK – The Cabinet Office and Number 10 begins implementing a specialist

rapid response unit to combat COVID-19 misinformation.
April 2020
2 Apr Google announces a range of new funding and initiatives intended to improve
COVID-19 news quality. 148 The Google News Initiative provides additional
support to First Draft for an online resource hub for journalists and to the
Crosscheck Network, which helps newsrooms respond to escalating harmful
content. In Australia, Google provides funding for media literacy research and
educational resources and works with the ABC to launch a COVID-19 news
bulletin on Google Assistant. 149
Google also gives US$5 million to support fact-checking programs, 150
particularly in Europe (focusing on the countries with the highest number of
COVID-19 cases) and in Spanish-speaking and Latin-American countries.
6 Apr Google implements new policies to mitigate misleading claims and promote
transparency on COVID-19 related information provided in apps available
through Google Play. 151 Only official government apps and verified health
apps which provide medical or support services related to COVID-19 or which
support local responses, may leverage COVID-19 related keywork searches in
the Google Play Store.
7 Apr Facebook limits forwarded messages in WhatsApp to reduce viral

misinformation. Once a message has been forwarded through a chain of 5 or
more chats, the message is labelled with a double arrow icon and may only be
forwarded to a single person at a time. 152
14 Apr Global – United Nations Communications Response initiative to combat the

spread of disinformation and misinformation is launched.
146
Facebook, Keeping People Safe and Informed About the Coronavirus.
147
Instagram, Keeping people safe, informed and supported on Instagram.
148
Google, $6.5 million to help fight coronavirus misinformation.
149
Google, Australian code of practice on disinformation and misinformation: Google Initial Report,
May 2021.
150
151
Sam Tolomei, ‘Google Play updates and information: Resources for developers’, Android Developers
Blog.
152
Casey Newton, WhatsApp puts new limits on the forwarding of viral messages.
acma | 103
21 Apr Twitter, in partnership with UNESCO, launches the #ThinkBeforeShareing
campaign, intended to raise awareness about conspiracy theories and
improve media literacy. 153
22 Apr Facebook announces modifications to provide location information for posts

shared by high-reach pages, to increase users’ ability to discern the ‘reliability
and authenticity’ of posts they see in their feeds.
28 Apr Republic of Ireland – BAI publishes CodeCheck: A Review of Platform

Compliance with the EC Code of Practice on Disinformation.
May 2020
6 May Facebook announces the membership of its Oversight Board. 154 The Board
was formed to make final and binding decision on whether specific content
should be allowed or removed from Facebook and Instagram. Users are able
to appeal Facebook moderation decisions to the Board to have their content
restored. Cases may also be referred to the Board directly by Facebook.
9 May Australia – Anti-lockdown protests in Melbourne.
11 May Twitter introduces additional labels and warning messages on tweets about
COVID-19 containing misleading information or disputed claims. Labels are
applied retrospectively, where false and misleading content has a ‘moderate’
propensity for harm. Labelled content is identified using proactive ‘internal
systems’ and via fact-checking partnerships. 155
19 May Global – WHO Member States pass Resolution WHA73.1, which recognises
that managing the ‘infodemic’ is critical to controlling the COVID-19 pandemic.
20 May YouTube publishes a COVID-19 Medical Misinformation Policy. 156 The policy
prohibits content that contradicts guidance by the WHO or local health
authorities on the treatment, prevention, diagnosis, transmission or existence
of COVID-19 or social distancing and self-isolation advice/rules. An exception
to the policy is provided for educational, documentary, scientific or artistic
content that contextualises such claims as false. This policy is subsequently
updated on 14 October 2020 to specifically address vaccines. Content that
contravenes the policy is removed and, after 3 contraventions, offending
channels are terminated. Google stated that this policy was developed out of
its ongoing work with global and local health authorities to ensure its policies
are accurately addressing risks of serious physical harm or death. 157
22 May Facebook makes location information available on posts shared by high-

reach pages.
153
UNESCO, European social media campaign to address disinformation on Covid-19 &
#ThinkBeforeSharing.
154
Nick Clegg, Welcoming the Oversight Board.
155
Yoel Roth and Nick Pickles, Updating our approach to misleading information.
156
YouTube, COVID-19 medical misinformation policy.
157
May 2021.
104 | acma
26 May US - Black Lives Matter protests commence, following the killing of George
Floyd in Minneapolis. Protests occur in more than 60 countries over the
coming weeks.
28 May US – President Trump signs the Executive Order on Preventing Online

Censorship, seeking reform of online platform protections.
28 May Facebook expands its policy of verifying the identity of users managing pages
with large audiences to also verify accounts generating viral posts in the US
that ‘have a pattern of inauthentic behaviour’. 158 The purpose of the policy
change was to provide additional transparency and information to Facebook
users. If an account chooses not to confirm its information, its content may be
disabled or have diminished reach.
30 May Australia – Anti-vax and anti-5G protests held in Sydney, Brisbane and
Melbourne.
June 2020
Facebook implements controls to enable US Facebook and Instagram users

to limit electoral, political and social issue advertising. Facebook already
requires that advertising for electoral, political and social issues (sensitive
topics that are heavily debated, may influence the outcome of an election or
result in/relate to existing or proposed legislation) are authorised and include
‘paid for by’ disclaimers. Facebook states it will be expanded to social issue
advertising in 2021. 159 An archive of political advertisements can be searched
publicly in Facebook’s Ad Library.
1 Jun Australia – NSW Court of Appeal upholds 2019 decision, which found news
media companies liable for defamatory comments posted by users on their
public Facebook pages.
1 Jun Twitter discloses 3 networks of accounts to its archive of state-linked

operations.
4 Jun Facebook commences labelling of content from state-controlled media

outlets. On 17 June, Facebook also starts blocks advertising from state-
controlled media outlets targeting people in the US.
6 Jun Australia – Black Lives Matter protests in Sydney.
20 Jun Australia – Victoria experiences COVID-19 second wave, and reintroduction

of state-wide lockdown restrictions. Restrictions ease on 22 November.
25 Jun Facebook introduces a new notification to alert users if a news article they are
going to share is more than 90 days old. The intention being to improve
context, in addition to the existing feature ‘context button’, which gives users
information on the source of an article. 160
158
Anita Joseph, Verifying the identity of people behind high-reach profiles.
159
Facebook, Facebook response to the Australian disinformation ad misinformation industry code, p. 8.
160
Facebook, Providing People With Additional Context About Content They Share.
acma | 105
26 Jun Australia – ACMA publishes Misinformation and news quality on digital
platforms in Australia: A position paper to guide code development. 161
July 2020
15 Jul Facebook updates its Facebook Coronavirus Information Center to include a

dedicated section on common myths (identified by the WHO) on COVID-19.
29 Jul US – ‘Anti-trust hearing’: Apple, Google, Facebook and Amazon appear

before US House Judiciary Committee.
31 Jul Google makes changes to its advertising policies, prohibiting websites that
accept Google advertising from publishing hacked materials, and to prohibit
advertisements that directly facilitate or advertise access to hacked material
related to political entities, including those distributed by a third party. 162 This
policy is rolled out globally in November 2020.
31 Jul Google announces that it will implement a new feature called ‘About this Ad’,
which will enable users to see the verified name of an advertiser. 163
Additionally, Google announces the release of a new tool called Ads
Transparency Spotlight (an alpha extension from the Chrome Web Store),
which provides detailed information about all advertising on Google Search.
August 2020
Facebook implements a number of changes to diminish the spread of harmful

content and extremism on its platforms, including: 164
> instituting a waiting period to prevent newly created Groups from being
recommended until quality can be assessed
> limiting the number of Group invites a user can send
> requiring mandatory post approval by administrators for Groups with low
‘integrity signals’
> turning off commenting on particular posts when hate speech is detected
> removing non-recommendable Groups that a user has joined as a factor
for new groups recommended to that user.
4 Aug Lebanon – Beirut explosion.
4 Aug Adobe, Microsoft and The New York Times publish a white paper on the
Content Authenticity Initiative. First announced in November 2019, the
initiative aims to develop an industry standard for digital content attribution. As
part of this work, Adobe is developing a system for creators and publishers to
embed visual and audio-visual content with attribution data, creating an
‘attribution trail’ as changes are made to pieces of content. 165
161
ACMA, Misinformation and news quality on digital platforms in Australia: A position paper to guide code
development, June 2021
162
Leah Nylen, ‘Google Announces steps to counter spread of hacked materials before election’, Politico.
163
Mike Schulman, ‘Updates on our work to improve user privacy in digital advertising’, Google Ads &
Commerce Blog.
164
165
The Content Authenticity Initiative: Setting the Standard for Digital Content Attribution was co-authored
by Adobe with BBC, CBC/Radio-Canada, Microsoft Corporation, The New York Times Company, Stanford
Center for Blockchain Research, Truepic, University of California, Berkley and WITNESS.
106 | acma
4 Aug Facebook pilots a new feature in WhatsApp that labels messages that have
been forwarded at least 5 times with a magnifying glass icon. Users can click
on this icon to launch a web search to help them verify the contents of the
message. 166
5 Aug TikTok announces a number of measures to address misinformation,

disinformation and interference with the 2020 US presidential election. 167
These aim to:
> prohibit manipulated content that may mislead users by distorting the truth
of events in a manner that could cause harm (this was in addition to a pre-
existing prohibition on content from disinformation campaigns)
> expand its fact-checking partnerships
> add an election misinformation option to the TikTok reporting feature to
enable easy reporting of content or accounts for review
> introduce an election information centre to connect users to authoritative
information
> work with the US Department of Homeland Security Countering Foreign
Influence Task Force to help stop the threat and dangers of foreign
influence on elections.
6 Aug Facebook removes a coordinated inauthentic behaviour network that

operates from multiple regions, including Australia, linked to the digital media
outlet Truthmedia. 168
6 Aug US – President Trump issues an Executive Order that would require TikTok to
sell or spin off parts of its business to continue operating in the US. This order
was revoked and replaced by a review of apps with ties to ‘jurisdiction of
foreign adversaries’ initiated by President Biden in June 2021.
7 Aug Facebook announces the winners of its call for ‘requests for proposals’ to
study misinformation and polarisation. 169
8 Aug WhatsApp releases new verification measures. 170
11 Aug Facebook adds 2 new fact-checking labels: Altered and Missing Content, and
updates its warning labels for Partly False and Missing Context ratings.
Facebook maintains partnerships with more than 70 fact-checking
organisations internationally. In Australia, Facebook partners with the
Australian Associated Press and Agency France Presse, which are certified
by the International Factchecking Network. 171
13 Aug YouTube makes changes to its policies to improve reliability of US election-

related news, 172 including to remove election-related content that violates its
166
Manish Singh, ‘WhatsApp pilots new feature to fight misinformation: Search the web’, TechCrunch.
167
Vanessa Pappas, General Manager, TikTok US, Combating misinformation and election interference on
TikTok.
168
Facebook, ‘July 2020 Coordinated Inauthentic Behaviour Report’, Facebook Newsroom’, 6 August 2020.
169
Alex Leavitt, Kathryn Grant, ‘Announcing the winners of Facebook’s request for proposals on
misinformation and polarization’, Facebook Research Blog.
170
171
Facebook’s submission to the Senate Select Committee on Foreign Interference through Social Media,
p. 10; Facebook, Response to the Australian disinformation and misinformation industry code, May 2021.
172
Leslie Miller, An update on how YouTube supports elections, YouTube Official Blog.
acma | 107
policies (such as hacked information), removing content that encourages
others to interfere with democratic processes, and labelling information about
US electoral candidates.
Instagram introduces new authenticity measures, requiring accounts to
confirm their information when Instagram identifies patterns of potentially
inauthentic behaviour.
18 Aug US – State of Emergency declared in California due to wildfires.
19 Aug Facebook expands its Dangerous Organisations and Individuals policy to

include measures to limit the spread of content on Facebook and Instagram
from Groups that do not meet the definition of a ‘dangerous organisation’,
such as QAnon and other US-based militia organisations. 173 Under the
expanded policy, Facebook may take the following actions:
> remove Facebook pages, Groups and Instagram accounts where
discussions of potential violence are identified
> limit recommendations for associated Facebook pages and Groups and
Instagram accounts
> rank content from these pages and Groups lower in the News Feed
(implemented 16 September 2020)
> reduce the visibility in the search function
> disable the related hashtag function on Instagram
> prohibit advertising, the selling of products or using Marketplace and
Shop, on content related to these movements (on 29 September 2020 this
was expanded to a prohibition against praising QAnon or militarised social
movements) 174
> prohibit identified Groups from using the service to fundraise.
22 Aug Australia – Northern Territory general election.
20 Aug Twitter expands its account labels to identify accounts for government
officials, including heads of state, and state-affiliate media to the 5 permanent
members of the UN Security Council. 175
31 Aug Facebook announces research program on the impact of Facebook and

Instagram on key political attitudes and behaviours during the US 2020
elections.
September 2020
1 Sep Microsoft announces 2 AI technologies to assist with identifying

deepfakes. 176 The first is the Microsoft Video Authenticator, which can assess
the likelihood that an image or video has been manipulated. The second is a
Microsoft Azure authenticity tool that allows content creators to add ‘digital
hashes and certificates’ to a piece of content. Microsoft also announces its
partnership with several media companies, including the BBC, CBC/Radio-
Canada and The New York Times, to test the authenticity tool as part of its
Project Origin initiative.
173
Facebook, An Update to How We Address Movements and Organizations Tied to Violence.
174
Ibid.
175
Twitter Support, Expanding our work to identify state-affiliate accounts.
176
Tom But and Eric Horvitz, New Steps to Combat Disinformation.
108 | acma
3 Sep Facebook announces it will label content which seeks to delegitimise the
outcome of the US election (including from campaigns or candidates) or call
premature victory. 177
Facebook introduces forwarding limits on Messenger to slow the spread of
viral misinformation. Users can only forward messages to 5 people or groups
at a time. 178
10 Sep EU – EC Staff Working Assessment of the EU Code of Practice on

Disinformation.
10 Sep Google makes changes to its auto-complete policies in Search to remove

false predictions about candidates in the US presidential election. 179
10 Sep Google announces its Intelligence Desk, which is a global team of analysists
set up to monitor news and events and assess how Google systems are
performing against evolving news ‘narratives’, such as in relation to
COVID-19. Google employs a number of systems to elevate authoritative
sources of information, including surfacing fact-checked information and
elevating original reporting on Google Search, providing information and
knowledge panels that highlight credibility on particular issues, using machine
learning to recognise authoritative sources as well as providing additional
safeguards during breaking news or crisis events, such as the Breaking News
shelf on YouTube, which appears on the YouTube home page when a
significant news event occurs and features authoritative sources of
information. Google states that it continues to ‘improve its systems across
Google and YouTube so that [it can] detect breaking news contexts (and crisis
situations) and optimise for elevating authoritative sources’ and reduce the
spread of misinformation. 180
16 Sep Facebook expands it ‘downranking’ to pages and Groups that have been
restricted but not removed.
17 Sep Facebook implements further changes to how it manages Groups to further

reduce the spread of misinformation. 181 The changes include:
> preventing administrators/moderators of Groups that have contravened
Facebook’s policies from creating any new Groups for a period of time
> requiring posts from Group members who have contravened Facebook’s
policies be approved by the Group administrator for 30 days
> archiving Groups without active administrators/moderators
> removing health-related Groups from Group recommendations
> removing Groups that share misinformation in contravention of
Fakebook’s Community Guidelines
> downgrading rankings of Groups that share false information (identified by
fact-checkers) in recommendations and News Feeds
> labelling content that has been reviewed by fact-checkers.
177
Facebook, New steps to protect the US elections.
178
Facebook, Introducing a Forwarding Limit on Messenger, 3 September 2020.
179
Google, Our latest investments in information quality in Search and News.
180
Google, Australian code of practice on disinformation and misinformation: Google Initial Report, May
2021.
181
Tom Alison, Our latest steps to keep Facebook Groups safe.
acma | 109
17 Sep Twitter increases account security protections for a designated group of high-
profile US election-related accounts. 182 The group included accounts relating
to the US Executive Branch and Congress, US Governors and Secretaries of
State, Presidential campaigns, political parties, and candidates with Twitter
Election Labels running for the US House, Senate, or for US state Governor,
and major US news outlets and political journalists. Other accounts could also
make use of the new measures, which include:
> requiring strong passwords
> password reset protection enabled by default
> 2-factor authentication strongly encouraged
> internal security measures increased, such as more sophisticated alerts of
suspicious activity.
17 Sep Twitters updates its Civic Integrity Policy to specifically address false and
misleading information ‘intended to undermine public confidence in an election
or other civic process’. 183 Under the policy, Twitter may remove or label tweets
that make false or misleading claims:
> about laws underpinning civic processes
> which may undermine faith civic processes
> about the outcome of a civic process with the intention of interfering with
the that outcome.
Tweets labelled under this policy will have their visibility reduced.
18 Sep Twitter releases its Coordinated Harmful Activity policy 184 that sets out the
framework by which it assesses whether a group, movement or campaign is
engaged in Coordinated Harmful Activity. The framework defines the relevant
categories of harm, physical, psychological and informational, consistent with
other policies in the Twitter Rules. Coordination is defined as either:
> technical coordination, which ‘refers to the use of specific detectable
techniques of platform manipulation to engage in the artificial inflation or
propagation of a message or narrative’ (all forms are prohibited under the
Twitter rules)
> social coordination, which ‘refers to on or off Twitter coordination among a
group of people to amplify or propagate a specific message’ (some forms
constitute a violation of the Twitter rules).
Under the framework, if Twitter identifies Coordinated Harmful Activity, it may
take a number of enforcement actions on tweets and accounts, including to
limit their visibility and suspend accounts primarily used for Coordinated
Harmful Activity. Under the Twitter rules more generally, tweets may be
removed that violate its policies and accounts may be suspended for severe
or repeated violations of the Twitter rules.
22 Sep TikTok launches an Asia-Pacific Safety Advisory Council, made up of

independent experts to advise on TikTok’s content moderation policies as well
as broader matters relating to trust and safety at TikTok. 185
23 Sep WHO issues a joint statement on managing the COVID-19 ‘infodemic’.
182
Twitter Safety, Improved account security during the 2020 US election.
183
Twitter Safety, Expanding our policies to further protect the civic conversation.
184
Twitter policy, Coordinated harmful activity.
185
Arjun Narayan Bettadopur, Introducing the TikTok Asia Pacific Safety Advisory Council.
110 | acma
25 Sep TikTok appears before Australian Senate Select Committee on Foreign
Interference through Social Media.
25 Sep Google announces that it will block election advertisements globally after 3
November 2020 until 7 days after the 2020 US Presidential election day. 186
This included advertisements that were explicitly election related (an
advertisement is considered election related if it mentioned a current state of
federal officeholder or candidate, political party, or ballot measure), any other
advertisements that reference federal or state elections, or advertisements
that run based on targeting election-related search queries.
29 Sep Facebook prohibits advertising that support militarised social movements and
QAnon.
30 Sep US – First US Presidential debate.
30 Sep Facebook implements new measures to limit the activities of QAnon. When
users search for certain hashtags related to child safety, they will be
redirected to credible information. QAnon content identified as false by third-
party fact-checkers will also be limited in News Feed, filtered from Explore and
hashtags on Instagram, and labelled with additional contextual material. 187
October 2020
1 Oct TikTok launches fact-checking in Australia with the global news agency
AFP. 188 TikTok has fact-checking partners across 8 markets that assist TikTok
to remove misinformation from its platform. Users are also able to report
information that they believe may violate its misleading content policies, using
the ‘misleading information’ category.
2 Oct US – President Trump tests positive for COVID-19.
6 Oct Facebook makes further updates to its Dangerous Individuals and

Organisations policy on QAnon, to remove any Pages, Groups, and Instagram
accounts representing QAnon even if they do not contain violent content.
Facebook states that it is proactively detecting content that may contravene
this policy, rather than relying on user reports. Facebook notes that the policy
change was linked to evidence that QAnon content was spreading other kinds
of misinformation (other than content promoting violence), leading to real
world harm, such as claims that the US west coast wildfires were started by
certain groups. 189
7 Oct Microsoft announces new initiatives to support journalism and local

newsrooms. 190 Microsoft has commenced 4 pilot programs in the US in which
it will provide direct funding, improve available technology, help build capacity
around technological transformation, expand news distribution and coordinate
knowledge-sharing on successful approaches to revenue and funding.
186
Sara Fischer, Scoop: Google to block election ads after Election Day, Axios.
187
188
Arjun Narayan Bettadopur, TikTok partners with fact-checking experts to combat misinformation.
189
190
Mary Snapp, New steps to preserve and protect journalism and local newsrooms.
acma | 111
Microsoft is also piloting the Protecting Journalists Pro Bono Program to
provide legal support to news organisations.
7 Oct Facebook announces measures to prevent misinformation in relation to the

US presidential election results. These include notifications at the top of
Facebook and Instagram as well as on the posts of candidates, and directing
users to its Voter Information Center. Facebook states that, in the event that a
candidate or party called premature victory before the election had been
called by a major media outlet, it intends to include additional information
about the counting process in warning labels. If the election result is
contested, Facebook states that it would include information about the
declared winner in a notification at the top of Facebook and Instagram.
8 Oct Twitter discloses 5 networks of accounts to its archive of potential information

operations (accounts of these networks suspended for violations of Twitter’s
policies).
9 Oct Twitter announced updates to its Civic Integrity Policy to mitigate possible
attempts to interfere with the outcome of the US Presidential election. The
updates prohibit all users from making claims about who won the election
before the outcome is called by an authoritative source. Under the policy, an
authoritative source refers to an announcement from US state election
officials, or a public projection from at least 2 national news outlets that make
independent election calls. The policy also prohibits tweets intended to incite
interference with the results. 191 Twitter also announces a series of other
measures to support its expanded policy, with the intention of increasing the
context of tweets and encouraging users to consider more thoughtfully
consider retweeting information. For example, users will be prompted to view
credible information before being able to retweet a tweet with a misleading
information label. 192
13 Oct Facebook launches a global policy prohibiting advertising that discourages

vaccination. 193 Under the policy, advertising that advocates a particular
position on the government’s policy on vaccines would not necessarily
contravene Facebook’s Community Standards, provided they include a ‘Paid
for by’ label.
14 Oct US – Hunter Biden’s hacked emails published by The New York Post.
14 Oct Facebook and Twitter limit spread of a New York Post article on Hunter
Biden.
14 Oct Google updates its COVID-19 medical misinformation policy for YouTube to
address vaccines. 194
191
Blog post by Vijaya Gadde and Kayvon Beykpour, Additional steps we’re taking ahead of the 2020
US Election.
192
Ibid.
193
Kany-Xi Jon, Supporting Public Health Experts’ Vaccine Efforts – About Facebook.
194
May 2021.
112 | acma
16 Oct Twitter makes changes to its Hacked Materials Policy to no longer remove
hacked content unless it is directly shared by hackers or those acting in
concert with them. 195 Under the policy, Twitter will label tweets to provide
context instead of blocking links from being shared on Twitter.
17 Oct New Zealand – General election.
Australia – ACT general election.
Twitter introduces new feature to its platform, prompting users before they
retweet or quote content that is labelled as containing misleading information.
18 Oct TikTok removes content and accounts that promote QAnon for violating its
Community Guidelines.
20 Oct Facebook temporarily suspends civic and political Groups being

recommended in the US. 196
21 Oct Facebook redirects searches on QAnon (and related issues) to credible

resources from the Global Network on Extremism and Technology (the
academic research network of the Global Internet Forum to Counter
Terrorism). 197
22 Oct Facebook’s Oversight Board officially begins to hear cases with the intention
of prioritising cases with the greatest potential for impact on users globally. 198
23 Oct TikTok updates its notification settings globally so that creators of content that
violates its policies will be notified of the policy they have contravened and
provide them with the ability to appeal the decision. 199
26 Oct Facebook stops accepting new political advertising the week before the US
presidential election.
27 Oct US – Section 230 of the Communications Decency Act 1996 (US) hearing:
Facebook, Google and Twitter appear before the US Senate Commerce
committee
30 Oct Facebook suspends political group recommendations in the US.
31 Oct Australia – Queensland state election.
November 2020
Europe – COVID-19 second wave
195
Tweet by Vijaya Gadde, Twitter Legal, Policy and Trust & Safety Lead.
196
Ryan Mac and Craig Silverman, ‘Facebook Quietly Suspended Political Group Recommendations Ahead
Of The US Presidential Election’, BuzzFeed.
197
198
Brent Harris, Oversight Board to Start Hearing Cases – About Facebook.
199
TikTok, Adding clarity to content removals.
acma | 113
Indonesia – Passes Ministerial Regulation No.5 (MR%) which requires
Private Electronic System Operators to register with the government and
make content accessible for monitoring. The regulation also enables the
government to order the removal of prohibited content, including content
which ‘promotes social anxiety and disrupts public order’. 200
3 Nov Google blocks election advertising in the US.
Facebook temporarily stops all social issue, electoral or political advertising in

the US.
4 Nov US – Presidential election day.

Allegations of ‘election fraud’ (including from President Trump and QAnon) are
widespread.
4 Nov Google displays election results from The Associated Press when users
search for election results. 201
5 Nov Facebook removes the ‘Stop the Steal’ Group under its Coordinating Harm
policy.
5 Nov Facebook is reported to (temporarily) modify its News Feed to prioritise

mainstream media content, in the days immediately following the US
Presidential election. 202
8 Nov UK – UK government agree on a set of measures with Google, Facebook, and

Twitter to address COVID-19 vaccine disinformation and misinformation and
to promote reliable information.
16 Nov Australia – COVID-19 outbreak in Adelaide and short-term lockdown.
18 Nov YouTube includes links to authoritative information on COVID-19 vaccines to

its COVID-19 information panels. Users who click on the links are taken to
authoritative third-party sources, like the Centers for Disease Control and
Prevention or the WHO. Information panels are a feature used by Google to
provide additional information and context to matters it identifies as prone to
misinformation. 203
25 Nov Google announces that its political advertising reporting functionality and
advertising library will become available in Australia. Election advertisements
in Australia are advertisements that feature a political party, current elected
officeholder or candidate for the House of Representatives or Senate. Election
advertisements do not include advertisements for products or services,
including promotional political merchandise, such as t-shirts or advertisements
run by news organisations to promote their coverage of political parties,
candidates, or current elected officeholders. The new function allows users to
see how much money parties and other groups spend on targeted election
advertisements.
200
Eduard Lazarus, ‘The authoritarian threat of Indonesia’s latest internet bill’, The Interpreter, 7 June 2021.
201
Amanda Storey, Following the 2020 US Election with Google.
202
Kevin Roose, Mike Isaac and Sheera Frenkel, ‘Facebook Struggles to Balance Civility and Growth’.
203
YouTube Help, Information panel giving topical context.
114 | acma
Google also introduces new policies for election advertising in Australia,
requiring that they be labelled with a ‘paid for by’ disclosure. Advertisers must
also go through a verification process. Political advertising is also subject to
Google’s other advertising policies. Google states that targeting election
advertising is permitted based on general geographic post code, age, gender,
and context. Targeting is not permitted based on political affiliation. 204
28 Nov TikTok announces it will partner with fact-checkers for the US presidential
election results, to help reduce discoverability of content that prematurely
claims victory in a race before results are confirmed by the Associated
Press. 205 TikTok also displays a banner on content with unverifiable claims
about premature declarations of victory.
December 2020
2 Dec Twitter expands its Hateful Conduct Policy to address content that
dehumanises people on the basis of race, ethnicity, or national origin. 206 Any
content that violates this policy is removed through proactive detection and
automatic moderation.
3 Dec Facebook announces that it will also apply existing policies to remove
misinformation that can lead to imminent, physical harm, or false claims about
COVID-19 vaccines. Facebook advised that, in enforcing this policy, it would
regularly update the types of false claims that it would remove under its
policies. 207
3 Dec EU – EC presents its Democracy Action Plan.
9 Dec YouTube starts applying its Presidential Election Integrity Policy, under which
content that alleges the outcome was the result of widespread fraud will be
removed or errors will be removed (this policy does not relate to elections
outside the US). 208 For the 2020 US presidential election, this policy applies to
content uploaded on or after the 9 December 2020.
10 Dec Google launches a new feature in the UK to provide information on and

surface a list of authorised vaccines in a user’s location. 209 This feature will be
rolled out as other jurisdictions begin offering the vaccine.
Google provides a $15 million ‘Ad Grant’ to the WHO to provide public
services announcements about the pandemic. 210 Google has also provided
$4.8 million in Ad Grants to the Australian Federal Government and
Department of Health and $48 million to Australian not-for-profits, 211 and
$250 million in Ad Grants to more than 100 government agencies throughout
204
Google, Australian code of practice on disinformation and misinformation: Google Initial Report, May
2021, p. 26.
205
Eric Han, Head of Safety, TikTok US, Supporting our community on Election Day and beyond.
206
Twitter Safety blog, Updating our rules against hateful conduct.
207
Kang-Xing Jin, Keeping People Safe and Informed About the Coronavirus – About Facebook.
208
Spam, deceptive practices, & scams policies – YouTube Help (google.com).
209
210
Ibid.
211
May 2021.
acma | 115
2020. Google also announces it will provide $1.5 million for the creation of a
COVID-19 Vaccine Media Hub by the Australian Science Media Centre. 212
14 Dec US – Joseph R. Biden confirmed as President-elect by the Electoral College.
15 Dec UK – Government publishes Online Harms White Paper response.

EU – EU Commission publishes the Digital Services Act and the Digital
Markets Act.
15 Dec TikTok updates its in-app coronavirus resource hub with commonly asked
questions and answers about COVID-19 vaccines from public health
experts. 213 These changes are made as part of broader updates to TikTok’s
policies to support well-being.
Facebook introduces notifications to alert people if they had interacted with a
piece of COVID-19-related content that had been removed under its
misinformation policies, explaining why it was false and linking to credible
information. 214
16 Dec Twitter expands policies to address misinformation about COVID-19

vaccines. 215 Twitter requires people to remove tweets that advance harmful,
false or misleading narratives about vaccines, such as false or misleading
claims that vaccines are used for harmful purposes, have adverse effects or
are unnecessary. From early 2021, Twitter may put warning labels on tweets
that spread potentially misleading information about vaccines. 216
18 Dec Australia – COVID-19 cluster and local lockdown in Sydney’s northern

beaches. Restrictions are lifted 10 January 2021.
23 Dec Twitter introduces an emergency search prompt, in collaboration with the

Australian Red Cross. Australian users who search for keywords, such as
bushfire, flooding and cyclone, are directed to the Australian Red Cross
Twitter account and resources. 217
January 2021
6 Jan US – Storming of US Capitol building, following the ‘Save America Rally’ in

Washington DC.
6 Jan Facebook temporarily restricts several Group features, including:

> requiring administrators to review and approve posts before they are
visible in certain Groups
> automatically disabling comments where a high rate of hate speech or
content that incites violence is detected in a Group
> using artificial intelligence to downrank content that is likely to contravene
Facebook policies.
212
213
Cormac Keenan, Head of Trust & Safety, TikTok, Refreshing our policies to support community well-
being.
214
Guy Rosen, An Update on Our Work to Keep People Informed and Limit Misinformation About COVID-19
– About Facebook.
215
Twitter Safety, COVID-19: Our approach to misleading vaccine information.
216
Twitter Safety, Updates to our work on COVID-19 vaccine misinformation.
217
Kara Hinesley, Twitter launches a search prompt with the Australian Red Cross.
116 | acma
Facebook suspends President Trump’s account for 24 hours.
7 Jan YouTube removes President Trump’s address on the storming of the Capitol,
including his description of the perpetrators as ‘special’ under its Presidential
Election Integrity Policy.
Facebook suspends President Trump’s Facebook and Instagram accounts.
8 Jan Twitter permanently bans President Trump from its platform.
TikTok removes President Trump’s speeches, which reiterate claims of a

fraudulent election.
Google removes the Parler app from the Google Play Store.
Apple follows suit the following day, removing Parler from the App Store on
9 January.
11 Jan Facebook removes content including ‘Stop the Steal’ under its Coordinating
Harm policy. 218
12 Jan Twitter updates its Civic Integrity Policy to increase enforcement measures,
including that repeated violations can result in permanent suspension of an
account. 219
16 Jan Facebook temporarily bans weapon accessories and protective equipment

advertising until 22 January 2021. 220
20 Jan US – Inauguration of Joe Biden as 46th US President.
21 Jan Facebook refers its decision to indefinitely suspend President Trump from
Facebook and Instagram to its Oversight Board for independent consideration
due to its significance. 221
25 Jan Twitter introduces ‘Birdwatch’ pilot program, which allows users to tag and
comment on information they think might be misleading. Twitter aims to make
the program available globally. 222
27 Jan Facebook permanently bans civic and political Groups from being
recommended and rolls this policy out globally.
February 2021
Google expands its collaboration with Defending Digital Campaigns.
218
Facebook, Our Preparations Ahead of Inauguration Day – About Facebook.
219
Twitter Safety, An update following the riots in Washington, DC.
220
Guy Rosen, Our Preparations Ahead of Inauguration Day – About Facebook.
221
Nick Clegg, Referring Former President Trump’s Suspension From Facebook to the Oversight Board –
About Facebook.
222
Twitter, Introducing Birdwatch, a community-based approach to misinformation, 25 January 2021.
acma | 117
Adobe, Microsoft, Arm, BBC, Intel, and Truepic form the Coalition for
Content Provenance and Authenticity (C2PA). The purpose of C2PA is to
create open technical standards for certifying the origin and evolution history
of digital media content as a means of addressing disinformation and
misinformation. C2PA brings together the Content Authenticity Initiative and
Project Origin (originally focused on online news distribution), which will
continue to engage with their respective sectors, under unified standards.
Facebook publishes a list of specific claims it considers to be misinformation

that could cause imminent, physical harm. 223 It also expands its
Misinformation and Harm policy to include false and misleading claims about
vaccines more generally, such as that vaccines cause autism or SIDS, or
other alarmist content related to vaccines.
TikTok launches a ‘Know the Facts’ campaign in Australia. 224
8 Feb Facebook, following consultation with the WHO and other health
organisations, expands the scope of its policy to remove false content on
COVID-19 vaccines on Facebook and Instagram, 225 such that Facebook and
Instagram Groups, pages and accounts that repeatedly share false claims (as
determined by third-party fact-checkers) could be removed.
8 Feb Facebook announces that it will grant $120 million in advertising credits to
health agencies (including in Australian federal and state agencies), NGOs
and UN agencies to provide people with information on the COVID-19 vaccine
and preventive health. Facebook also makes modifications to its search
function to return higher instances of validated vaccine information and reduce
identified misinformation.
16 Feb Facebook temporarily bars Australians from finding or sharing news on the
platform, in response to the Australian Government’s News Media and Digital
Platforms Bargaining Code. The ban is lifted a week later on 23 February.
17 Feb Facebook and Instagram permanently ban celebrity chef Pete Evans for
repeatedly sharing misinformation about COVID-19 and vaccines.
17 Feb Twitter expands its account labels to identify accounts for government
officials, including heads of state, and state-affiliate media for G7 nations. 226
Twitter states that it intends to expand this policy. 227
22 Feb Australia – DIGI launches the Australian Code of Practice on Disinformation

and Misinformation and announces the 6 initial code signatories.
Australia – COVID-19 vaccine rollout commences.
223
Facebook, Facebook response to the Australian disinformation ad misinformation industry code, p. 26.
224
TikTok, Australian Code of Practice on Disinformation and Misinformation: Initial Report, May 2021.
225
Margaret Harding McGill and Sara Fischer, ‘Facebook says it will crack down on COVID vaccine
misinformation’, Axios.
226
Twitter Support, Expanding our work to identify state-affiliated accounts.
227
Twitter, Twitter: Australian Code of Practice on Disinformation and Misinformation Initial Report, p. 18.
118 | acma
23 Feb Twitter discloses 5 networks of accounts to its archive of potential information
operations (accounts of these networks suspended for violations of Twitter’s
policies).
25 Feb India – Introduces the Information Technology (Guidelines for Intermediaries

and Digital Media Ethics Code) Rules requiring social media and digital
streaming companies to remove content deemed unlawful.
March 2021
1 Mar TikTok introduces a new feature, allowing users to decide which comments
appear next to a post. This feature is in addition to existing controls that
enable users to filter spam and offensive comments, and specific keywords.
The new feature is also complemented by a prompt to TikTok’s Community
Guidelines, and suggests reconsidering posting a comment, when TikTok
detects that a comment may be ‘inappropriate or unkind’. 228
1 Mar Twitter updates its COVID-19 policy to include an enforcement strike system,
designed to help to educate the public on its policies, and further reduce the
spread of potentially harmful and misleading information on Twitter,
particularly for repeated moderate and high-severity violations of its rules. 229
4 Mar Facebook implements controls enabling Facebook and Instagram users to

limit electoral, political and social advertising in 90 countries. 230 In Australia,
the feature is available for political- and election-related advertising only. 231
13 Mar Australia – Western Australia state election.
15/16 Facebook expands on its 15 December 2020 notifications by also applying

Mar information labels to content discussing COVID-19 vaccines on Facebook and
Instagram. 232 The labels include credible information from the WHO on the
safety and efficacy of COVID-19 vaccines. Facebook also launches a
COVID-19 Information Center on Instagram on 16 March. 233
17 Mar Facebook implements further restrictions to Groups, including that Groups

that have contravened the Community Guidelines will appear lower down in
recommendations (Groups that repeatedly contravene Facebook’s Community
Guidelines are removed). 234 Facebook will notify people when they are joining
a Group that has violations and limit invite notifications for these Groups.
Where Facebook identifies ‘severe harm’, it will remove Groups without
implementing gradual enforcement.
18 Mar Australia – NSW floods start. Floods subside by the end of March 2021.
228
Tara Wadhwa, ‘New Tools to promote kindness on TikTok’, TikTok Safety, 11 March 2021.
229
Twitter Safety, Updates to our work on COVID-19 vaccine misinformation, 1 March 2021.
230
Naomi Cleit, Launching the largest voting information effort in US history.
231
Availability for ads about social issues, elections or politics | Facebook Business Help Centre.
232
Facebook, ‘Mark Zuckerberg announces Facebook’s plans to help get people vaccinated against
COVID-19’, Facebook newsroom, 15 March 2020.
233
Instagram, ‘Helping people stay safe and informed about COVID-19 vaccines’, Instagram
Announcements, 16 March 2021.
234
Tom Alison, Changes to keep Facebook Groups Safe.
acma | 119
18 Mar Apple News features special curated section on the flood emergency. 235
24 Mar Instagram introduces a range of updates to improve COVID-19 information

accuracy, including only recommending COVID-19 related accounts if they
belong to a credible health organisation.
31 Mar Facebook introduces a new feature to give users greater ability to control
comments on their posts. 236 The change is available to all users but will
enable news media organisations to more effectively moderate comments for
potentially defamatory content.
April 2021
Early India – Beginning of COVID-19 second wave.

Apr
Twitter introduces a timeline prompt to link Australian users with the

Australian Government Department of Health’s COVID-19 landing page
providing information on vaccines and vaccination. 237
8 Apr US – Facebook, Google and Twitter appear before the House Committee on
Energy and Commerce on the role and responsibility of digital platforms in
promoting extremism and misinformation following the 2020 US election.
12 Apr Australia – Australian Government updates official health advice regarding

the AstraZeneca COVID-19 vaccine (Pfizer preferred for under 50s).
13 Apr Facebook-funded report on media literacy in Australia launched by the

Western Sydney University, Queensland University of Technology and the
University of Canberra. 238
14 Apr Facebook’s Oversight Board starts accepting appeals from Facebook and
Instagram users that want content removed from these services. Previously,
users were only able to appeal to have content restored. 239
19 Apr Facebook announces proactive measures to limit real-world harm arising

from the Derek Chauvin verdict on the murder of George Floyd. Facebook
indicates that, as temporary emergency measures, it may limit the spread of
content that its systems predict will contravene its policies, and remove events
organised in temporary, high-risk locations that contain calls to bring arms. 240
26 Apr Facebook permanently removes Craig Kelly MP’s Facebook and Instagram
accounts for repeated violations of its COVID-19 misinformation policy.
235
Apple, Apple News: Australian Code of Practice on Disinformation and Misinformation Initial report.
236
Josh Taylor, Facebook now lets users and pages turn off comments on their posts, The Guardian, 31
March 2021.
237
Twitter, Twitter: Australian Code of Practice on Disinformation and Misinformation Initial Report, p. 8.
238
Facebook, Facebook response to the Australian disinformation ad misinformation industry code.
239
Oversight Board, The Oversight Board is accepting user appeals to remove content from Facebook and
Instagram, April 2021.
240
Monica Bickert, Preparing for a Verdict in the Trial of Derek Chauvin, Facebook Newsroom, 12 April
2021.
120 | acma
A Facebook spokesperson noted: ‘We don't allow anyone, including elected
officials, to share misinformation about COVID-19 that could lead to imminent
physical harm or COVID-19 vaccines that have been debunked by public
health experts. We have clear policies against this type of content and have
removed Mr Kelly's Facebook Page for repeated violations of this policy’. 241
27 Apr TikTok announces it will open a European Transparency and Accountability

Centre, similar to its US Transparency Center. Independent experts will be
able to observe how TikTok implements its content moderation polices. 242
29 Apr Facebook temporarily hides posts containing #ResignModi due to some posts
violating Community Standards. Facebook notes this was a mistake and
restores posts with this hashtag after several hours. 243
May 2021
1 May Australia – Tasmanian state election.
5 May Facebook’s Oversight Board upholds Facebook’s decision to suspend

President Trump from Facebook and Instagram but determines that indefinite
suspension is inconsistent with Facebook’s normal penalties. The Board
insists that Facebook review the matter within 6 months to apply a
‘proportionate response that is consistent with the rules that are applies to
other users of its platform’. 244
Facebook announces its response to the Oversight Board finding on 4 June,
noting that it will suspend former President Trump’s account for 2 years,
ending 7 January 2023. The suspension is the highest penalty under a new
enforcement protocol developed by Facebook in response to the Oversight
Board’s findings. Facebook will also provide more information in its
Transparency Center about when it applies its ‘newsworthiness allowance’
and allows content to stay on the platform on public interest grounds, where it
might otherwise contravene Facebook’s policies. 245
12 May UK – UK Government publishes its draft Online Safety Bill. This includes
a proposal for independent regulator Ofcom to establish an advisory
committee on disinformation and misinformation and publish periodic
reports on this issue.
22 May Australia – DIGI publishes the code commitments and transparency reports
of all 8 signatories to the code. Facebook signals its intent to expand its
fact-checking capability in Australia in the second half of 2021. 246
241
Jade Macmillan and Brett Worthington, Facebook removes Craig Kelly's page, says former Liberal MP
breached misinformation policies, ABC News, 26 April 2021.
242
Natasha Lomas, TikTok to open a ‘Transparency’ Center in Europe to take content and security
questions, TechCrunch, 27 April 2021.
243
Kari Paul, Facebook blocked hashtag calling for Narendra Modi to resign over pandemic, The Guardian,
29 April 2021.
244
Oversight Board, Case decision 2021-001-FB-FBR, Board Decisions, 5 May 2021.
245
Nick Clegg, In response to oversight board, Trump suspended for two years; will only be reinstated if
conditions permit, Facebook Newsroom, 4 June 2021.
246
Josh Machin, Facebook response to the Australian disinformation ad misinformation industry code,
Facebook Australia, 21 May 2021.
acma | 121
26 May WhatsApp brings a lawsuit against the Indian Government over new laws that
give the Indian Government powers to monitor online activity. 247
EU – European Commission publishes its Guidance on Strengthening the

Code of Practice on Disinformation, recommending an expansion of scope.
Facebook places further limits on users who repeatedly share false content
that has been fact-checked, by reducing the distribution of all their posts in
News Feed, rather than just reducing reach of individual posts. 248 Facebook
also lifts ban on claims that COVID-19 is man-made. 249
June 2021
4 Jun Nigeria indefinitely suspends Twitter following its deletion of a tweet by

President Muhammadu Buhari and announces plans to prosecute any
Nigerians defying the government’s ban. 250
247
Hannah Ellis-Petersen, WhatsApp sues Indian government over ‘mass surveillance’ internet laws, The
Guardian, 26 May 2021.
248
Facebook, Taking Action Against People Who Repeatedly Share Misinformation, Facebook Newsroom,
26 May 2021.
249
Alex Hern, ‘Facebook lifts ban on posts claiming Covid-19 was man-made’, The Guardian, 28 May 2021.
250
Associated Press, Nigeria’s 40 million Twitter users banned from site as government-enforced
suspension takes effect’, ABC News, 7 June 2021.
122 | acma
Appendix D: International
regulatory approaches
Several jurisdictions have introduced or are considering anti-disinformation or
misinformation initiatives relating to digital platforms.
To inform the development of this report, the ACMA has monitored a range of
international regulatory developments and consulted with international regulators such
as Ofcom, the Broadcasting Authority of Ireland, and the Canadian Radio-television
and Telecommunications Commission, to share common experiences and learn from
their approaches.
Europe
The Australian Government’s response to the DPI asked digital platforms to draw on
learnings from the EU Code to inform the development of the Australian code(s).
The EU Code is a voluntary, outcomes-based code that sets out a list of high-level
commitments and principles that signatories agree to follow to protect users from
disinformation. The code was introduced October 2018 as a pioneering self-regulatory
code to address disinformation. Current signatories include Facebook, Google,
Microsoft, Twitter, Mozilla, TikTok, and several advertising industry groups.
Since its inception, there have been significant developments to evaluate and redress
the EU Code. An overview of these developments is listed below and a timeline is
provided at Figure 21.
acma | 123
Figure 21: Timeline of key EU Code developments
Code of Practice
European Code of Practice on Disinformation (October 2018)
↓
Self-assessment reporting
First annual reports from code signatories (October 2019)
↓
Key reports feeding into European Commission’s assessment
ERGA report of the European Commission’s
EC’s summary & analysis VVA Independent Study
Implementation of the EU 2019 Elections
of self-assessment reports on the Implementation of
Code of Practice Reports
(October 2019) the Code (May 2020)
(May 2020) (June 2020)
Formal assessment of the EU Code Self-assessment reporting

EC Staff Working Assessment of the EU Code of COVID-19 disinformation monitoring programme
Practice on Disinformation (September 2020) (July 2020 to present)
↓
Public consultations Stakeholder engagement
Consultation for guidance Consultation on political Stakeholder discussions with platforms, ad industry
on revised EU Code ad transparency trade associations, academics, fact-checkers, ERGA,
(April 2021) (January to April 2021) EDMO, EU Member States (early 2021)
Guidance from European Commission

Guidance on strengthening the Code of Practice on Disinformation (May 2021)
↓
Revised Code of Practice
Revised European Code of Practice on Disinformation (expected end of 2021)
↓
Proposed upcoming legislation and strategic plans
Legislative measures for
Digital Services Act Digital Markets Act European Democracy
political advertising
(expected 2023) (expected 2023) Action Plan
(Q3 2021)
124 | acma
European Commission’s assessment of the EU Code
In September 2020, the European Commission published its formal assessment of the
EU Code for its initial 12-month period from October 2018 to October 2019. 251 The
assessment report is a ‘staff working document’ and does not contain any formal
recommendations for the European Commission or code signatories.
The assessment found that the code is a valuable instrument and provides a
framework for a structured dialogue between relevant stakeholders to ensure greater
transparency of platforms’ policies against disinformation in the EU.
Some of the key successes of the EU Code identified by the European Commission
include:
> greater transparency of political advertising, including clearly labelling political ads,
identifying candidates, political parties, and the creation of ad libraries to contain
all political ads served
> platforms’ removal of fake accounts, malicious bots, and spam content
> greater investment in technology to give prominence to trustworthy information
sources and make it easier for users to critically assess and find diverse
perspectives about topics of public interest through new tools
> new services and features for users to flag potential instances of disinformation
and be warned about content that has been fact-checked and rated as false or
misleading
> new policies and tools that provide researchers and the fact-checking community
with better access to platform data.
Weaknesses of the code were mostly attributed to limited participation, lack of

independent oversight, and the need for further data to be provided from platforms to
the research community and relevant regulatory authorities. Key shortcomings include:
> lack of collaboration of stakeholders, including platforms, fact-checkers and
advertisers
> no data showing the effectiveness of news tools to increase user engagement with
trustworthy information sources or enhance critical thinking
> no user-friendly and uniform procedure available on all platforms for users to flag
possible disinformation cases and be adequately informed about the outcome of
their actions
> inadequate approaches to raise users’ awareness of content labelled as false or
misleading, including when a user has previously come across this content
> lack of information on the integration of trustworthiness indicators in platforms’
algorithms.
The European Commission also identified some emerging areas in the disinformation
environment that remain unaddressed in the code, including:
> manipulative online behaviours, such as fake engagement techniques aimed at
inflating the perceived popularity of certain content
> micro-targeting of political advertising
> KPIs and data for monitoring the code to address the lack of common reporting
structures and allow for cross-platform comparisons.
251
European Commission, Commission Staff Working Document: Assessment of the Code of Practice on
Disinformation – Achievements and areas for further improvement, September 2020.
acma | 125
COVID-19 disinformation monitoring program
At the onset of the COVID-19 pandemic, the Commission asked platforms to report
monthly on the effectiveness and impact of their policies in relation to the COVID-19
pandemic. These reports included detail on blocking accounts that engage in vaccine-
related disinformation, advertisers disseminating COVID-19 and vaccine-related
misinformation, and efforts to increase engagement with fact-checking organisations.
In January 2021, the European Commission provided some advice to assist

signatories with their reporting. Signatories were advised to provide more data on the
evolution and spread disinformation during this time, the granular impact of their
actions at the level of EU countries, and to have a greater focus on vaccine-specific
disinformation. Monthly reporting requirements are expected to continue until at least
the end of 2021.
Strengthening the EU Code

On 26 May 2021, the European Commission issued formal guidance setting out how
platforms should step up their measures to address the shortcomings identified in the
assessment of the EU Code.
Some of the key recommendations from the guidance included:

> an expansion of the code to include misinformation in some areas, as well as
private messaging services
> wider participation from both established and emerging platforms, as well as fact-
checkers, content assessment organisations, and technology developers
> a requirement for signatories to publicly justify their reason for opting out of certain
code provisions
> enhanced code provisions on the demonetisation of disinformation on advertising
channels.
It also suggests an enhanced monitoring and reporting framework, which includes the
development of:
> platform-specific and industry-wide KPIs
> a harmonised reporting template to enable cross-platform comparisons
> publicly accessible transparency centres created by platforms
> a new data access framework for the research community
> a dedicated taskforce consisting of representatives from signatories, EDMO,
ERGA and other relevant experts, to evolve and adapt the code to technological,
market, and legislative developments.
The Commission expects a first draft of a revised EU Code in the European

autumn 2021.
Digital Services Act

The European Commission has proposed a Digital Services Act (DSA) to enable
users, consumers, and businesses to continue to operate and grow with digital
developments and challenges in Europe. 252 Its goal is to create a safer digital
environment in which the fundamental rights of all users of digital services are
protected. It is not anticipated that the DSA will be enacted before the end of 2022.
252
European Commission, The Digital Services Act package, December 2020.
126 | acma
The DSA provides a series of new, harmonised obligations for digital services in the
EU. Some of these include:
> rules for the removal of illegal goods, services, or content online
> safeguards for users whose content has been deleted by platforms
> obligations for platforms to take risk-based action to prevent abuse of their
systems
> transparency measures for online advertising and platform algorithms
> facilitating access for researchers to key platform data, and public access to
independent audit reports of very large online platforms’ risk mitigation measures.
Services’ obligations under the DSA are graduated on the basis of their size and reach
in the EU. It is proposed that arrangements will be overseen by EU Member States,
and for some larger platforms, the European Commission directly.
The DSA proposes establishing a co-regulatory backstop, which would include a

strengthened EU Code and a more robust framework for monitoring the code.
The proposal also covers new advertising rules that will empower users in
understanding and making informed decisions about the ads they see. Users will be
clearly informed whether and why they are being targeted by each ad and who paid for
the ad. They should also see clearly when content is sponsored or organically posted
on a platform.
As a complementing piece to the measures proposed in the DSA, the European

Democracy Action Plan was presented by the European Commission in December
2020 and contains a strategic plan specific for countering disinformation and its impact
on democracy in Europe. The action plan proposes improving the existing EU’s
toolbox for countering foreign interference, such as new powers that allow costs to be
imposed on bad actors. The European Commission will increase funding to support
new innovative projects to fight disinformation and promote media literacy, including
those by civil society organisations and academic institutions.
United Kingdom
Online Harms White Paper and draft online safety bill
The UK government’s Online Harms White Paper puts forward a regulatory framework
that assesses potentially harmful content in a proportionate manner and is based on
evidence of risk of harm. The proposed approach aims to improve transparency for
users about which content is and is not acceptable on different platforms and will
enhance users’ ability to challenge the removal of content where this occurs.
In December 2020, the UK government released its preliminary response of the

consultation on the Online Harms White Paper. 253 It appointed Ofcom as the
independent regulator that will be responsible for protecting users from online harms.
The white paper has informed the development of the UK’s May 2021 draft Online
Safety Bill. 254 The Bill applies to a range of user-to-user and search services that are
categorised based on their online presence, functionality, and high-risk features.
253
UK Department for Digital, Culture, Media & Sport, Online Harms White Paper – Initial consultation
response, December 2020.
254
UK Department for Digital, Culture, Media & Sport, Draft Online Safety Bill, May 2021.
acma | 127
Category 1 services, likely to include Facebook, TikTok, Instagram and Twitter, will
need to mitigate the risk of harmful and illegal content on their services, including
disinformation. They will also need to conduct and publish assessments of the steps
they have taken to address any adverse effects cause by their platforms. Ofcom will
publish guidance to assist providers to comply with their obligations to carry out risk
assessments.
Category 1 services will also have a statutory duty of care to safeguard UK users’
access to journalistic content share on their platforms. Under this duty of care,
services will need to consider the importance of journalism when undertaking content
moderation, have a fast-track appeals process for journalists’ removed content, and
will be held accountable for the arbitrary removal of journalist content. Articles by
recognised news publishers shared on Category 1 services will be exempt.
Ofcom will establish and maintain an advisory committee specific to addressing

misinformation and disinformation and will advise on how:
> regulated services should deal with such content on their services/encountered in
or via search results
> it will use its transparency reporting powers in relation to misinformation and
disinformation
> it will use its media literacy functions to counter such content.
The committee will publish a report within 18 months and maintain subsequent
periodic reporting.
COVID-19 response
In March 2020, the Cabinet Office and Number 10 began implementing a specialist
rapid response unit to combat false and misleading narratives about COVID-19, to
ensure the public has access to credible and authoritative information to protect
themselves. 255
The government has also agreed on measures with Google, Facebook, and Twitter to
limit the spread of disinformation and misinformation regarding COVID-19 vaccines. 256
Together, these platforms have agreed to:

> ensure a timely response to disinformation and misinformation content flagged to
them by the government
> continue to work with public health bodies to ensure that authoritative messages
about vaccine safety reach as many people as possible
> participate in policy forums to improve responses to disinformation and
misinformation and prepare for future threats.
France
France adopted laws in late 2018 against the ‘manipulation of information’ (Les enjeux
de la loi contre la manipulation de l’information), which apply to online platforms with
more than 5 million monthly French users, including social media sites and search
engines. 257
255
UK Government, Government cracks down on spread of false coronavirus information online, March
2020.
256
UK Department for Digital, Culture, Media & Sport, Social media giants agree package of measure with
UK Government to tackle vaccine disinformation, November 2020.
257
Les enjeux de la loi contre la manipulation de l’information 2018.
128 | acma
The laws encourage platforms for take measures such as:
> improve transparency of algorithms
> promote content from credible sources
> fight against accounts that disseminate fake information
> provide users with further insight into the source of content, especially
advertisements
> provide users with information on how the content is being disseminated
> promote media literacy.
The national media regulator, Conseil supérieur de l'audiovisuel (CSA), has formal
information-gathering powers to oversee digital platforms, including the ability to
conduct performance reports and issue recommendations to platforms. It can also
request certain data points from platforms and broader information requests such as
providing more insight into the transparency of algorithms.
Platforms are required to submit annual performance reports to CSA outlining the
measures they have taken to fight against disinformation in France. The first reports
were supplied in May 2020 and CSA released its first evaluation report in July 2020.
As an EU Member State, France’s national regulatory response is operated in parallel

with the EU Code framework.
Germany
Germany also has national level arrangements to complement the EU Code. While
there is no general law that prohibits the creation and dissemination of disinformation,
there are several legal provisions that may be applicable to safeguard individuals or
the public from disinformation on digital platforms.
In 2017, Germany passed the Network Enforcement Act, which aims to combat hate
speech and misinformation on social media platforms that have more than 2 million
registered users in Germany. 258 Under the Act, social media platforms have 24 hours
after receiving a user complaint to remove any content that is ‘clearly illegal’, as
defined by provisions of the Criminal Code that cover the dissemination of
propaganda, encouragement of serious violent offenses, the incitement of crime and
hatred, among others.
Social media platforms must also offer their users a complaints mechanism that is
accessible and transparent. Any decision made on a user’s complaint must be
communicated to the user and any affected user in a clear and timely manner.
Germany has also recently updated its interstate media treaty for broadcasting
regulation. The treaty contains updated and expanded journalistic standards that will
provide a legal basis for regulation of online services within its broadcasting
framework. Article 19 of the Act seeks to protect free expression within the bounds of
journalistic standards.
258
Act to Improve Enforcement of the Law in Social Networks (Network Enforcement Act, NetzDG) - Basic
Information 2017.
acma | 129
Ireland
As an EU Member State, Ireland’s response to disinformation has primarily fallen
under the framework of the EU Code.
The Broadcasting Authority of Ireland (BAI) is the national regulatory authority for
Ireland in relation to the EU Code, and has played a leading role in the ERGA sub-
group on disinformation.
In April 2020, BAI published its report CodeCheck: A Review of Platform Compliance
with the EC Code of Practice on Disinformation as its contribution to ERGA’s
assessment. 259 The report outlined the responses by the platforms under Pillars A and
C and examines in-depth specifically which of the actions occurred within Ireland
under Pillar D and E. It found that significant progress was made by digital platforms.
However, there remains weaknesses in the content of the code and procedures for
reporting, monitoring, and enforcing the commitments, particularly at a national level. 
The Irish Government is currently drafting the Online Safety and Media Regulation
Bill. It is expected that the proposed legislation will address the regulation of harmful
online content. BAI will be dissolved, and existing personnel transferred to a new
Media Commission. An Online Safety Commissioner with specific responsibility for
overseeing the regulatory framework for online safety will also be established within
the Media Commission. 260
United States
Under Section 230 of the US Communications Decency Act, providers and users of an
‘interactive computer service’ are provided a general immunity from civil liability arising
from third-party content that is deemed obscene, excessively violent, harassing, or
otherwise objectionable. Under this Act, digital platforms are treated as distributors –
not the publisher or speaker – of any information provided by a user of the service.
There has been recent interest in whether actions taken by platforms against
misleading content threatens freedom of expression. Under the former Trump
administration, there were discussions about reforming Section 230, including an
Executive Order on Preventing Online Censorship that sought to clarify whether
platforms’ actions constitute editorial conduct and therefore run against the principles
underlying the immunity from liability.
The Biden administration has yet to announce a formal position on changes to section
230. However congressional hearings continue to examine the impact of the
legislation. In March 2021, chief executives from Facebook, Alphabet (Google) and
Twitter were asked to testify to the House Committee on Energy and Commerce
relating to the role and responsibility of digital platforms in promoting extremism and
misinformation following the 2020 US election.
Canada
Canada is undertaking ongoing efforts to modernise its legislative and regulatory
framework for communications.
259
Broadcasting Authority of Ireland, New report highlights inconsistencies across digital platforms in
tackling disinformation, April 2020.
260
Mondaq, Ireland: Irish Media and Broadcasting Law Update, May 2021.
130 | acma
The Canadian government’s 2020 report Canada’s Communications Future: Time to
Act acknowledged the need to address online misinformation and recommended
legislative reform to address the spread of harmful content on digital platforms. 261
During the COVID-19 pandemic, Canada’s federal government has been underscoring
the importance of trusting experts and consulting credible sources. It has invested
significantly in projects that facilitate public awareness tools and online workshops to
help citizens become more resilience and think critically about COVID-19
disinformation. 262
The Canadian Department of Heritage has also proposed new measures to address
online harms. Legislation would see the creation of a new regulatory body to
implement rules for online speech and assess and enforce compliance with new
regulations. The new regulator would be given auditing powers over platforms’ content
moderation and be able to implement a 24-hour takedown notice regime to help
reduce the spread of potentially harmful content online.
The Canadian government is also leading a Diversity of Content Online Working

Group with members from Australia, France, Germany, Finland, as well as from the
private sector and civic society. In June 2021, a new set of guiding principles were
announced to foster greater exposure to diverse news and information, bolster
resilience to disinformation and misinformation, and promote greater transparency of
the impacts of algorithmic treatments of online content.
The principles are intended to guide a range of stakeholder actions, including those of
digital platforms. Signatories have agreed to develop specific commitments by
December 2022 to implement the principles.
Singapore
The Protection from Online Falsehoods Manipulation Act (POFMA) allows government
ministers to order news outlets, internet service providers, digital platforms and users
to include warnings that their pages or posts contain false statements and include links
to government fact-checking websites. 263
The Infocomm Media Development Authority (IMDA), a statutory board under the
Minister for Communications and Information, is responsible for administering POFMA
and has invoked it more than 50 times, primarily against independent media or people
who have criticised the government or its policies. 264 The POFMA Office, situated
within IMDA, issues directions and notices upon the instructions of ministers and
monitors and enforces compliance with these directions.
Critics have said that POFMA is a threat to free speech and have highlighted the
inherent conflicts of interests in laws that give the government extensive powers and
discretion to decide what constitutes misinformation. 265 Digital platforms have also
raised concerns, saying that POFMA is a censorship tool, and activists and social
261
Recommendation 94: ‘We recommend that the federal government introduce legislation with respect to
liability of digital providers for harmful content and conduct using digital technologies … we also encourage
the federal government to continue to participate actively in international fora and activities to develop
international cooperative regulatory practices on harmful content’.
262
Government of Canada, Supporting Canadians to Think Critically About Online Health Information,
April 2020.
263
POFMA Office, Protection from Online Falsehoods and Manipulation Act (POFMA), 2021.
264
Human Rights Watch, Singapore: ‘Fake News’ Law Curtails Speech, 2021.
265
La Trobe University, Fighting Fake News: A Study of Online Misinformation Regulation in the Asia
Pacific, Carson A et al, 2021.
acma | 131
groups are fearful that it is being used for political gain. The Singapore government
maintains that the law only tackles falsehoods and that legitimate criticism and free
speech are not affected.
POFMA has also had a controversial impact on Singapore’s online political

advertising, with rules requiring internet intermediaries to keep records of all online
political advertising content. This has prompted Google to ban all such content in
Singapore.
Malaysia
Malaysia has recently adopted the Emergency (Essential Powers) (No. 2) Ordinance
2021. The ordinance makes it a criminal offense to create, publish and disseminate
‘fake news’ that is likely to cause fear or harm to the public; or to fail to take down such
material upon the government’s request. ‘Fake news’ is defined as ‘news, information,
data and reports which is or are wholly or partly false relating to COVID-19 or the
proclamation of emergency’. It draws heavily from the repealed Anti-Fake News Act
2018 and is effective until 1 August 2021.
The ordinance has been criticised for failing to establish standards for determining
what is false, raising the risk that it could be used to silence criticism or other content
that government may not like. 266 It also allows for criminal punishment, regardless of
whether the offending individual or company had a prior understanding of the content
being false, misleading, or deceptive.
Taiwan
Similar to the EU code, major digital platforms such as Google, Facebook and
Yahoo and other local industry groups in Taiwan have signed a voluntary, self-
regulatory code to address concerns about false information on digital platforms. 267
Much of the reporting framework from the EU has been translated into the Taiwan
code, including platforms periodically reviewing the results of their activities and
proactively continuing to establish dialogue with third parties and government
agencies to support and maintain transparency.
India
In February 2021, the Information Technology (Guidelines for Intermediaries and
Digital Media Ethics Code) Rules commenced. The rules are overseen by the Ministry
of Electronics and Information Technology and apply, among others, to publishers of
news and online curated content, and intermediaries that enable the transmission of
such content.
Social media platforms will need to appoint a grievance officer who acknowledges
complaints within 24 hours and resolves them within 15 days. They will also be
required to set up a robust complaints-handing mechanism and publish a monthly
compliance report detailing the complaints received and any subsequent action taken.
266
Human Rights Watch, Malaysia: Revoke ‘Fake News’ Ordinance, March 2021.
267
Central News Agency (CNA), Five major players such as Facebook, LINE to prevent false information
take the lead in self-discipline (translated), June 2019.
132 | acma
Certain platforms will be required to provide details about the origin of pieces of
content and messages that has been identified as offensive by the Indian government.
This extends to private messaging services that offer end-to-end encryption for their
users. Some experts have expressed concern that this approach may supress
freedom of expression and increase censorship by digital platforms.
acma | 133
Appendix E: Other Australian
Government initiatives
There are several government initiatives that intersect with the operation and
administration of the code and its intended policy objectives. Many of these initiatives
result from recommendations of the ACCC’s Digital Platforms Inquiry (DPI).
This appendix provides public information about some of the initiatives currently
underway.
Government responses to the Digital Platforms Inquiry

News Media and Digital Platforms Mandatory Bargaining Code
The news media bargaining code was developed following a recommendation of the
ACCC’s DPI. It received royal assent on 2 March 2021. The purpose of the news
media bargaining code is to support the sustainability of the Australian news media
sector by addressing the imbalance in bargaining power between digital platforms
and Australian news businesses. It provides a framework for designated digital
platforms and registered news business corporations to make commercial agreements
regarding the availability of news on digital platform services. The obligations under
the code are enlivened when a digital platform is designated by the Treasurer and the
ACMA has registered a news business.
Under the news media bargaining code, designated platforms must develop proposals
to recognise original news content when they make it available and distribute that
content. This intersects with Objective 4 of the code, which places obligations on
digital platforms to enable users to make more informed choices about the source of
news and factual content accessed via their platforms.
On 3 March 2021, the ACMA opened applications for Australian news businesses to
register under the news media bargaining code. At this stage, no digital platforms have
been designated. Both Google and Facebook have struck commercial arrangements
with a number of news outlets voluntarily following the passage of the legislation. The
ACCC has also granted interim authorisation for Country Press Australia (CPA)
members to collectively negotiate with Facebook and Google over payments for their
news content that appears on these platforms.
Improving digital media literacy in the community

The Department of Infrastructure, Transport, Regional Development and
Communications (DITRDC) is responsible for implementing Recommendation 12 of
the DPI, to establish a network of experts to develop media literacy materials. DITRDC
is focusing on adult media literacy, with particular attention to vulnerable and Culturally
and Linguistically Diverse (CALD) communities. The work is in early stages of
development, having been reprioritised due to the impacts of the COVID-19 pandemic.
Currently, it is looking at opportunities to work with the already established Australian
Media Literacy Alliance, which comprises a range of members, including the ABC,
universities, the National Film and Sound Archive, Museum of Australian Democracy
and Australian Library and Information Association.
Online Privacy Code and review of the Privacy Act

The Attorney-General’s Department (AGD) is progressing reforms to strengthen
privacy protections online by introducing a new binding online privacy code for social
media and online platforms that trade in personal information and increasing penalties
134 | acma
and enforcement measures. The online privacy code will require organisations to be
more transparent about the handling of personal information, stop using or disclosing
an individual’s personal information upon request, and follow stricter rules about
handling personal information of children and other vulnerable groups. An exposure
draft will be released shortly for public consultation.
As industry will be invited to develop the online privacy code following passage of the
legislation, DIGI may have a role in code development. If an industry code developer
cannot be identified, the new binding code would be developed by the OAIC.
Additionally, in response to recommendations in the DPI, AGD is conducting a review

of the Privacy Act 1988. The AGD has been considering submissions received in
response to its issues paper, which outlined the current law and sought feedback on
potential issues relevant to reform. The AGD has also conducted targeted consultation
with stakeholders, including other government departments and agencies, state and
territory government departments, private sector entities, stakeholder representative
organisations and peak bodies, and international governments and privacy regulators.
A discussion paper analysing stakeholder feedback and seeking submissions on
reform proposals is expected to be issued in 2021. Feedback received through
submissions to that discussion paper, and further consultations will inform the review’s
final report for government. Some of the issues being considered by the review include
whether the requirements for notice to collect personal information should be
strengthened, whether individuals should be able to object to the collection of their
personal information, and a potential right to erasure of personal information
External dispute resolution

Following Recommendations 22 and 23 of the DPI, DITRDC is leading a process to
understand the existing dispute resolution mechanisms of digital platforms. The
process involves mapping the internal dispute resolution practices of major digital
platforms and gathering data from consumers and businesses to better understand
their concerns. DITRDC is consulting with industry and government stakeholders to
inform the development of policy advice.
There may be some similarities between this initiative and section 7.4 of the code,
which obliges signatories to establish a facility for addressing non-compliance by
signatories with the code.
ACCC digital platform services inquiry 2020–25

On 10 February 2020, the government, in response to the recommendation of the DPI,
directed the ACCC to conduct an inquiry into markets for the supply of digital platform
services, including search engine services, social media services, online private
messaging services, digital content aggregation platform services, media referral
services and electronic marketplace services. The ACCC must report to the Treasurer
every 6 months until 31 March 2025.
The ACCC published its first interim report on 30 September 2020. The report
examined online private messaging services in Australia and updated the ACCC’s
previous analysis of search and social media platforms. In this interim report, the
ACCC observes that the terms and conditions of online private messaging services
allow for the collection of a broad range of information about users, but do not provide
clarity on how that data will be used. The ACCC also found that standard terms
provided to small businesses seeking to advertise on large digital platforms are
potentially unfair. The report continues to advocate for Recommendations 22 and 23
of the DPI regarding internal dispute resolution mechanisms and the establishment of
an ombudsman scheme to resolve complaints and disputes with digital platforms.
acma | 135
The ACCC published its second interim report on 28 April 2021. This report provided
in-depth consideration of competition and consumer issues associated with the
distribution of mobile apps to users of smartphones and other mobile devices. Among
other things, the ACCC found that Apple and Google could do more to prevent and
remove apps that feature subscription traps and other scams. The report also found a
need for better redress and dispute resolution for consumers harmed by such apps.
The ACCC continues to support Recommendations 22 and 23 of the DPI. The ACCC
considers that these mechanisms could cover complaints by both third-party app
developers and mobile app users to help address identified deficiencies.
ACCC digital advertising services inquiry

On 10 February 2020, the government directed the ACCC to conduct an inquiry into
markets for the supply of digital advertising technology services and digital advertising
agency services. On 28 January 2021, the ACCC published its interim report. The
interim report does not make any preliminary findings, but seeks feedback on a
number of proposals, including that industry develop a voluntary standard to enable
full, independent verification of services provided by demand-side platforms (for
example, Google Ads and Display & Video 360). This proposal seeks to address
opacity in the supply chain that prevents users from making informed decisions about
the use of Google’s advertising services.
The ACCC is due to provide a final report to the Treasurer by 31 August 2021.
Other government initiatives

Online Safety Act
The Online Safety Bill 2021 (Online Safety Bill), which passed the Senate on 22 June
2021 and is expected to receive royal assent in the coming weeks, provides for the
development of a set of basic online safety expectations for social media services,
relevant electronic services and designated internet services. The core expectations
include that the provider of the service will:
> take reasonable steps to ensure that end-users are able to use the service in a
safe manner
> provide reporting and complaints mechanisms for end-users
> provide specified information to the eSafety Commissioner on request.
The Online Safety Bill also empowers the eSafety Commissioner to require the
providers of these services to prepare either ad hoc or periodic transparency reports
about their compliance with the basic online safety expectations during a specified
timeframe.
In addition, the Online Safety Bill provides for the establishment of a complaints and
take-down scheme for serious cyber-abuse material targeting Australian adults on
social media services, relevant electronic services and designated internet services.
This is material that an ordinary reasonable person would conclude is likely intended
to have an effect of causing serious harm to an Australian adult, and which an ordinary
reasonable Australian adult would regard as being menacing, harassing or offensive in
all the circumstances.
The updated online content scheme within the Online Safety Bill provides for the
removal of seriously harmful material in certain circumstances. It reflects and clarifies
the current regime in Schedules 5 and 7 of the BSA. Under the scheme, the eSafety
Commissioner’s take-down powers for class 1 content (such as child sexual
exploitation material and pro-terrorist material) are expanded to reach content that is
hosted overseas. The eSafety Commissioner is also empowered to require services
136 | acma
provided from Australia restrict certain class 2 content (which maps to R18+ material
under the National Classification code) for users under the age of 18. In addition, 8
sections of the online industry (including digital platforms represented by DIGI) are to
work collaboratively to develop new industry codes to address class 1 and class 2
content. The Bill provides that the eSafety Commissioner should make reasonable
efforts to ensure that an industry code is registered within 6 months of the Act
commencing. The eSafety Commissioner would also have the power to create
industry standards.
Defamation law reform

In November 2004, Attorneys-General endorsed the Model Defamation Provisions
(MDPs). The states and territories then enacted legislation to implement the MDPs
within their jurisdictions, collectively referred to as the National Uniform Defamation
Law. A process to review Australia’s National Uniform Defamation Law is currently
underway. This review is being led by Attorneys-General and is being progressed in
2 stages.
Stage 1 is now largely complete, with Attorneys-General endorsing the Model

Defamation Amendment Provisions (MDAPs) in June 2020. State and territory
Attorneys-General have agreed that the MDAPs will commence within their
jurisdictions on, or as soon as possible after, 1 July 2021. The Stage 2 review is
underway and is on the 2021 agenda for the Meeting of Attorneys-General.
Part A of Stage 2 focuses on the responsibility and liabilities of digital platforms for
defamatory content published online. Part A seeks to ensure that the MDPs are fit-for-
purpose in the digital age, noting the 1 June 2020 decision by the NSW Court of
Appeal to uphold a 2019 decision, which found news outlets were liable as ‘publishers’
for defamatory comments posted by third parties on their public Facebook pages.
Part B of Stage 2 considers if defamation law is discouraging reports of misconduct to
employers, police and other investigative or disciplinary bodies. Although the code
does not address defamatory comments explicitly, Stage 2 defamation law reforms are
consistent with a broader policy focus and regulatory push for digital platforms to be
held more accountable for the content on their services.
Voluntary transparency reporting protocols

The Department of Home Affairs is leading Australia’s engagement with industry,
governments, academia and civil society to develop the Voluntary Transparency
Reporting Framework under the auspices of the Organisation for Economic
Cooperation and Development (OECD). The framework, co-funded by Australia, seeks
to establish a common standard for online platforms to implement regular and
transparent public reporting on the steps they are taking to prevent, detect and remove
terrorist and violent extremist content on their platforms.
The third (and final) phase of the project commenced in May 2021 and is expected to
be finalised later this year. Once established, the framework may be used as a
standard of mandatory transparency reporting to the government for companies’
responses to ‘Abhorrent Violent Material’ under the proposed Online Safety Bill.
Australia is supportive of the OECD’s plan to renew negotiations in 2022 to develop a
more substantive Framework 2.0, which would be aimed at larger online platforms and
services and will better reflect their capabilities and resources.
Department of Foreign Affairs and Trade ongoing counter-disinformation work

The Department of Foreign Affairs and Trade (DFAT) has established a counter-
disinformation branch that works in collaboration with other agencies to monitor,
analyse, assess and respond to disinformation that is contrary to Australia’s national
interests. It also engages with its overseas partners, including through the Australia-
acma | 137
United States Ministerial Consultations (AUSMIN) Working Group on Disinformation. A
key element of this work is building resilience and limiting the spread of disinformation
through capacity building in the Indo-Pacific region, and developing international
norms to counter disinformation.
Home Affairs ongoing counter-disinformation work

In response to the threat of COVID-19 misinformation, disinformation and scams
targeting Australians at a time of global crisis, Home Affairs established an All Source
Fusion Cell (ASFC) to identify, analyse, assess and make recommendations for action
to counter COVID-19 manipulated information activity and monitor key themes and
trends in manipulated narratives.
The ASFC draws on information provided by departments and agencies across

government and from open-source platforms to produce fused summary reports.
Between March and June 2020, the ASFC produced over 60 reports, and made over
180 referrals to digital industry and law enforcement for further prevention, disruption
and strategic communications action to minimise and counter the spread of
misinformation and disinformation, and to protect Australians from scams.
Home Affairs has also been consulting with relevant departments and agencies to
explore options for government to address malicious uses of the deepfake software,
such as through the generation of disinformation, misinformation and malinformation.
Online Media Elections Protocol

The Electoral Council of Australia and New Zealand (ECANZ) is currently developing
an Online Media Elections Protocol in consultation with online platforms. The
protocol seeks to put in place arrangements between online platforms and electoral
bodies at the state, territory or commonwealth level to deal with content that potentially
breaches electoral legislation or the terms and conditions of online platforms.
The protocol makes references to relevant portions of the code. The code requires
signatories to develop and implement policies that provide users with greater
transparency about the source of political advertising carried on digital platforms.
Department of Health misinformation initiatives

The Department of Health (DoH) has a strategy to counter misinformation and works
with Department of Home Affairs and state and territory departments, as well as a
number of specific advisory groups (for example, the Aboriginal and Torres Strait
Islander Advisory Group on COVID-19 and the Culturally and Linguistically Diverse
Communities COVID-19 Health Advisory Group). As part of its strategy, DoH
proactively engages with platforms via liaisons to promote reliable evidence-based
information sources such as the DoH and World Health Organization.
DoH also monitors its own social media pages and accounts for misinformation and
will report misinformation to digital platforms directly via the platform reporting
functions or through platform liaisons. DoH also has social listening tools for
uncovering broad trends, monitoring sentiment and emerging themes, and identifying
popular hashtags. These social listening tools monitor external accounts to collect
information, which is used to develop new content to address common themes.
Digital Economy Strategy 2030

On 6 May 2021, the Prime Minister, the Treasurer and the Minister for
Superannuation, Financial Services and the Digital Economy released Australia’s
Digital Economy Strategy. The strategy sets out how Australia will secure its future as
a modern and leading digital economy by 2030. It builds on the government’s existing
138 | acma
data and digital initiatives, sets out further action under the 2021–22 Budget and
defines future pathways to 2030.
The Digital Economy Strategy builds on the government’s previous investments,

including the Digital Business Plan, Cyber Security Strategy 2020 and Australia’s Tech
Future. Development of the strategy was led by the Digital Technology Taskforce in
the Department of the Prime Minister and Cabinet, in close consultation across
government and with industry, academia, and non-government organisations.
Australia’s Cyber Security Strategy 2020

The Australia’s Cyber Security Strategy 2020, developed by the Home Affairs in
conjunction with PM&C, commits $1.67 billion over 10 years to Australian cyber
security initiatives. These include enhancing critical infrastructure and governmental
networks, new laws to enhance privacy, consumer and data protection, blocking
threats automatically and acting against cyber-attacks, as well as a voluntary code of
practice for the Internet of Things (released on 30 September 2020).
acma | 139
Appendix F: Development of key
performance indicators
The code is an outcomes-based, self-regulatory instrument. As articulated in the
ACMA’s position paper, an outcomes-based regulatory model has 3 distinguishing
features:
1. Regulation is drafted as high-level outcomes or objectives that must be met.
2. Regulated entities develop their own systems to achieve the outcomes specified in
the regulation.
3. Regulated entities are required to demonstrate delivery of these outcomes to the
regulator, with enforcement and compliance measures in place should a failure to
achieve an outcome occur.
The success of this approach hinges on signatories being able to demonstrate – both
to DIGI as the code administrator and the wider community – how they are meeting
each of their nominated outcomes under the code. Measures taken by signatories
under the code should be capable of being monitored and measured over time,
supported by a robust and transparent reporting framework.
As a requirement under the code, signatories have until 22 August 2021 to agree on a
format for future annual reports and a guideline that will inform the data and other
information to be included in subsequent reports. To promote consistency in reporting,
we found that a uniform report format should be developed, requiring all code
signatories to identify their services covered by the code, measures to address the
outcomes and the specific metrics or KPIs they will use to measure success under
each outcome.
This appendix outlines the ACMA’s views and recommendations on measurement and
reporting. Signatories may wish to consider this as guidance ahead of their next
annual reporting process.
Measurement framework
KPIs are an important method to measure how effective an entity is performing against
its objectives. KPIs need to be intimately connected to code commitments and be
measurable through high-quality sets of data.
Data could be sourced through a variety of methods. The most valuable data is likely
to come from signatories directly. Signatories may choose to use additional sources,
such as consumer or academic research.
The ACMA has suggested that KPIs be separated into 2 tiers, based on whether they
relate to the signatory’s service(s), the industry, or the misinformation environment
more broadly.
Tier 1 – Signatory-specific KPIs

A signatory publishes its annual report outlining how it will meet its commitments and
what KPIs it will use to demonstrate performance under the code. These KPIs are
monitored and reported on in each subsequent annual report.
The aim of signatory-specific KPIs is for signatories to demonstrate how their specific
business will comply with the code. As each signatory can determine how they meet
140 | acma
their code obligations under an outcomes-based framework, signatory-specific
reporting is necessary to demonstrate the performance of measures.
A signatory’s annual reporting should cover both their progress towards implementing
specific measures and the extent to which their measures have been successful in
meeting the overarching objectives of the code. It is important that signatories are able
to distinguish between the measures (for example a new tool or policy) and how they
are to be assessed.
Examples:
> A signatory has introduced a new tool that allows its users to view more details
about who is funding advertising on its service. As part of its annual reporting, the
signatory could report on the number of times this tool has been used by
Australian users. This would include data directly from the signatory and include
month-by-month increases or decreases in use.
> A signatory has partnered with a local fact-checking organisation to independently
review and tag content on their service. The signatory could report on the
operation of the fact-checking organisation and provide quantitative data points
such as how many pieces of content were tagged, how many accounts/pages and
pieces of content tagged were Australia-specific, and the average time taken to
review and tag content as false or misleading. Other data could include
engagement with posts before and after being fact-checked.
> A signatory has introduced a new tool that enables greater detection of bot
activity. The signatory could report on any month-to-month or year-to-year
changes in bot detection since the tool was introduced, how many bots discovered
were being operated in Australia, or the average time for the signatory to identify
and remove any bot accounts.
> A signatory has implemented a new misinformation-related policy. It could outline
why this policy is being implemented and demonstrate how it was communicated
to users (such a through emails and/or notifications of changes to terms of
service/users guidelines), and how the signatory will measure the success of this
policy.
A signatory may have built a COVID-19 information and resource centre for its users
to access during the pandemic. It could provide data on the number of Australian
users who accessed this, a list of the official health links provided and the number of
times Australian users clicked these links, and a list of the most common search terms
used.
Tier 2 – Industry-wide and industry-group KPIs

DIGI, in collaboration with code signatories, should develop a series of cross-industry
KPIs that align to the broader objectives of the code. This data allows for a better
understanding of how each platform is performing under the code, as well as how
signatories are performing as an industry in Australia.
Industry-wide and industry-group KPIs include broader metrics than signatory-specific

KPIs and enable cross-platform comparisons.
These KPIs would need joint agreement by code signatories, be generalisable across
all signatories and/or different groups of signatory-types (such as social media sites,
search engines, or news aggregators), and directly align to the overarching objectives
of the code. Given the breadth of signatories’ service types, not all KPIs would apply to
all signatories.
acma | 141
One of the main criticisms of the EU Code is the lack of KPIs to allow cross-platform
comparison. As outlined in Appendix D, recent guidance from the European
Commission provides that the EU Code should be strengthened with clear KPIs. This
allows for a measure of both the impact of actions taken by signatories, as well as the
overall impact of the EU Code on Disinformation in Europe. As no industry-wide KPIs
were supplied in signatories’ initial annual reports, we therefore strongly encourage
the inclusion of these in the next code reporting process.
Signatories may choose to include these as part of their individual annual reports.
Alternatively, they could report these metrics to an appropriate repository, such as
DIGI or the code sub-committee, which could use this data in its oversight capacity.
DIGI could report these metrics to the ACMA or publicly at an aggregated, industry-
wide level so as not to call out any specific signatory.
Examples:
> All signatories could agree to report annually on the number of Australian users
who have made a report about harmful misinformation on their service(s) covered
by the code, and the percentage of total reports that resulted in an action being
taken by the platform. Signatories may provide high-level figures about whether
the problem was addressed.
> Signatories that offer social media services could demonstrate how they have
addressed the same piece of content circulating on multiple platforms, such as a
deepfake video or verifiably false information from a public figure. This could show
similarities and differences in approaches and how efficiently certain content is
addressed.
> Signatories that offer search engine services could identify the number of sites
and pieces of content that have been identified as disinformation and
subsequently ‘buried’ in search algorithms.
142 | acma

Australian Government Plans New Media Censorship Laws

Uploaded by

Copyright:

Available Formats

Australian Government Plans New Media Censorship Laws

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Australian Government Plans New Media Censorship Laws

Uploaded by

Copyright:

Available Formats

A report to government on the adequacy of

digital platforms’ disinformation and news

T +61 2 6219 5555

All other rights are reserved.

Written enquiries may be sent to:

4. Assessment of the code 48

5. Assessment of platform performance 64

6. Considerations for future reform 76

Appendix A: Full list of recommendations and findings 85

Appendix B: Signatory assessment reports 90

Appendix C: Timeline of key events 98

Appendix D: International regulatory approaches 123

Appendix E: Other Australian Government initiatives 134

Appendix F: Development of key performance indicators 140

The online propagation of disinformation and misinformation presents an

Digital platforms have introduced a range of measures in response to the

Digital platforms have come together to develop a single outcomes-based code

The code adopts an outcomes-based regulatory approach that allows a range of

The code also provides signatories flexibility to implement measures to counter

Our assessment identifies further improvements that should be made to the

We also consider improvements to the code should be made in relation to:

The code includes commitments to establish administrative functions within 6 months

The ACMA recommends a number of actions by government to bolster industry

Given these circumstances, continued monitoring is required and the ACMA

Initial signatory reports identify challenges in obtaining relevant data on platform

There are also opportunities for improved collaboration between government

The ACMA makes 5 recommendations to the government in its report.

Recommendation 1: The government should encourage DIGI to consider the findings

Recommendation 3: To incentivise greater transparency, the ACMA should be

Recommendation 5: In addition to existing monitoring capabilities, the government

Defining the issue

Online disinformation and misinformation are relatively novel and dynamic

At a high level, many distinguish disinformation from misinformation on the

From a regulatory perspective, the ACMA recommended in its position

The ACMA thanks representatives of the platforms, DIGI, broader industry,

Through an examination of false, misleading or unproven narratives arising from the

2.1. Exposure and susceptibility

Most Australians experienced misinformation about COVID-19 in 2020. When asked

High Low None Don't Know

Figure 2: Experience of COVID-19 misinformation, by main source of news (%)

High Low None Don't Know

This research relies on respondents both knowing and accurately self-reporting on

Figure 3: Experience of misinformation, by misinformed groups (%)

High Low None Don't Know

Additionally, throughout 2020, many Australians reported relying on social media as a

Wave 1 (April 2020) Wave 2 (Dec 2020/Jan 2021)

A lack of trust in authoritative sources does, however, appear to be an indicator of

informed misinformed informed misinformed informed misinformed informed misinformed

Trust Neither Distrust

In follow-up focus group discussions undertaken by the N&MRC, several parents,

Finding 2: Access to authoritative and trusted sources of news and information is

2.2. Role of digital platforms

Table 1: Weekly digital platform usage, by general usage, and consumption of

# Platform General usage Consumption of COVID-19

News consumption by platform users

Figure 6: Consumption of news and information about COVID-19, among

Total Consumption No consumption

Active consumption Incidental consumption

I don't go to Facebook to look at information, I just go to socialise.