Written evidence submitted by Full Fact [MEL 240]

About Full Fact

1. Full Fact fights bad information. Our mission is to build a better information environment to restore trust. Our team of independent fact checkers, technologists, researchers, and policy specialists tackle the harm done by misinformation in four ways. We check claims made by politicians, public institutions, in the media and online. We ask people to correct the record where possible to reduce the spread of specific claims. We campaign for system changes to help make bad information rarer and less harmful, and we advocate for high standards in public debate.
2. Full Fact is a registered charity. We're funded by individual donations, charitable trusts, and by other funders. We have received funding from both Meta and Google. Details of our funding can be found on our website. Full Fact’s expertise covers public debate, misinformation and disinformation.

Summary

3. Full Fact welcomes the Committee’s inquiry. We are grateful for the opportunity to provide evidence: this focuses on the questions about democratic engagement and modernisation of elections.
4. We are concerned that the Representation of the People Bill fails to address the risks of misinformation and disinformation. Due to the scope of the Bill, key amendments relating to media literacy and online safety have been rejected, removing from Parliament the opportunity to address or debate the urgent risks to our democracy from online platforms.
5. During the 2024 election campaign, Full Fact carried out more than 450 hours of monitoring, while our AI tools analysed over 136 million words in 142,909 articles, transcripts and social media posts. We produced approximately 217 verdicts on claims or repeated claims, and published over 150 pieces of website and video content.
6. Analysis of Full Fact’s 2024 election fact checking revealed that despite pre-election fears based on the election experiences of other countries, the course of the 2024 campaign was not ultimately defined by sophisticated deepfakes, but rather by traditional political spin and by misinformation narratives circulating on social media platforms, which were then disseminated by political figures themselves.
7. Full Fact published six fact checks about the 2026 Gorton and Denton by-election. This included claims about candidates as well as those made by politicians or parties, including the misleading use of bar charts - a regular feature of election campaigns, in our experience of more than a decade.
8. During the run up to the 2026 May elections, Full Fact is using in-house automated monitoring tools to automatically flag to our fact checking team when any candidates share claims we have already debunked. We are also running images through SynthID to identify AI-generated content being shared by candidates. We have also expanded the sources we monitor to include TikTok and YouTube shorts. We would be happy to follow up this written evidence with further relevant evidence based on our monitoring and fresh fact checking after the election campaign.
9. Full Fact will publish a report in June focusing on growing uncertainty about what to trust and the impact this has on democracy. It explores how political and technological change is reshaping the way information is produced and distributed, and what this means for the effectiveness of our laws and institutions. We are happy to provide a copy of this report to the Committee in due course.

Democratic engagement and young people’s participation

10. Full Fact and Internet Matters recently published a joint analysis of polling conducted in November 2025. Lowering the voting age creates an opportunity to strengthen democratic participation, but only if young people are supported to navigate political information online confidently, including misinformation and manipulated content.
11. Children said they are navigating political content well before voting age: 74% of those aged 13-14 have seen content about news, politics or current affairs online. Children also reported that they lack foundational skills for evaluating political information: Only 53% of young people aged 13-17 who have seen political information online are confident in telling whether information online about politics is true or false, and just 59% feel confident distinguishing fact from opinion online. This was echoed by parents, who said they think children are unprepared to make informed electoral decisions: 52% of parents think young people are unprepared to vote, with only 49% expressing confidence in their child’s ability to recognise satire.
12. Young people were asked to assess four real-world political claims from the 2024 General Election campaign. Each example was designed to test different aspects of evaluating political information, including identifying fact versus opinion, recognising satire, interpreting claims based on official data, and considering potential political bias. Uncertainty was common. Overall, 41% of young people selected “I don’t know / unsure” for at least one of the four examples. This pattern was consistent across age groups, with 13-year olds just as likely to say they were unsure (45%) as 16-year-olds (47%). On average, 24% of young people selected “I don’t know / unsure” for each claim.
13. The polling also identified low trust in election campaigns and political actors from a young age: 63% of young people say they are concerned about voters being misled by false or misleading claims during elections. 60% said they are concerned that AI-generated content may affect the results of a general election. Alongside these concerns, many young people report low trust in political actors and messaging. Over 60% say they tend to ignore what political parties and politicians say because they do not know if they can trust them. This view is evident from an early age, with 57% of 13-to 14-year-olds agreeing with this statement.
14. Young people believe there is shared responsibility for helping them to identify false or misleading information online, including across schools, parents and carers, government, and social media companies. The polling found that young people also express strong support for interventions that would help them navigate political information more confidently. The majority (79%) say that social media platforms should be required to label or fact-check false or misleading information, and 75% believe schools should do more to teach young people how to evaluate political information. Taken together, these findings point to an expectation that improved education and greater platform responsibility should work in tandem.

Recommendations: what the Government could do to improve democratic engagement and young people’s participation

●        Schools must be supported to strengthen media and digital literacy across the curriculum through access to high-quality resources and comprehensive teacher training. A recent independent review of the curriculum in England highlighted the need to equip young people with the ability to make informed decisions and to help them to understand how opinions, AI-generated content and satire can all influence democratic participation.

●        The government must commit to sustained funding to deliver media literacy education outside schools, including for the Electoral Commission to deliver evidence-based public information campaigns on issues such as misinformation. The Communications and Digital Committee highlighted that media literacy provision outside schools is patchy and often reliant on underfunded third sector organisations, limiting its reach and consistency. This gap matters. Internet Matters research shows that many young people rely on trusted adults when navigating online information: 52% say they turn to a trusted adult when they encounter distressing news content on social media, and 51% say they would speak to a trusted adult if they wanted to verify whether news content is accurate.

●        The Government should amend the Representation of the People Bill to give social media companies a statutory duty to support users’ media literacy on platforms, including labelling AI generated content, design features that support critical evaluation (e.g. read-before-you-share prompts and source information labels), and user controls for recommender systems. Appendix 1 contains a draft amendment for this, which was rejected on the basis that it was deemed beyond the Bill’s scope.

The role of online misinformation and disinformation in recent UK elections: examples

15. In the run up to the UK general election in July 2024 there was a flurry of warnings that the campaign could be dominated by deepfakes that could undermine democracy, and widespread concerns about the potential for AI technology to deceive and mislead. 70% of MPs polled in a YouGov survey prior to the 2024 election were concerned about AI-generated content increasing the spread of misinformation and disinformation in the run up to polling day.
16. Research from the Centre for Emerging Technology and Security found 16 viral AI-enabled disinformation cases in the UK, leading the paper’s author to say that “there was no conclusive evidence that such tools had affected election results.” Full Fact’s work also showed that the reality was more nuanced, with a blend of old-fashioned political spin, online misinformation on social media, and rudimentary, easily debunked “cheapfakes”.
17. However, there has been a marked increase in the use of AI-generated content in our fact checking work, and a series of high-profile deepfakes in recent years. In November 2024, Full Fact suspected AI involvement in four published fact checks; in October 2025, this figure had risen to at least 27. This reflects a wider pattern observed by other fact checking organisations across Europe. Even where deepfakes do not directly influence the outcome of an election, they fuel wider mistrust in institutions and in information itself.
18. There were examples of apparently synthetic content that couldn’t be verified. An audio clip circulated purporting to be then-shadow health secretary Wes Streeting swearing and claiming he didn’t care about Palestinians being killed in the Israel-Gaza war. An audio clip supposedly of Keir Starmer claiming that he hates Liverpool was widely circulated online, with one post having received over 400,000 views as of 1 July 2024. We were not able to determine whether the Starmer clip was generated with AI, cleverly edited or was simply the work of a skilled impersonator. But we did not see any evidence to suggest it was real, and we identified versions of the clip that had been circulating since October 2023.
19. Female politicians were also targeted by deepfake pornographic smears, with the psychological damage such content caused potentially leading to a ‘chilling effect’ on the willingness of other women to enter politics.
20. "Cheapfakes"—less technically advanced pieces of fabricated content that are obviously false—proved effective in misleading voters. Full Fact investigated a video of Rachel Reeves pausing for several seconds after being asked about public finances under a Labour government, was clipped and shared with captions like, “Cat got your tongue, Rachel?”—implying that she had been caught off guard or unprepared. However, a review of the full footage revealed technical glitches during the interview that caused a delay between the question and her response. Another viral image showed Rishi Sunak standing in front of a Morrisons supermarket sign, with part of the logo obscured to spell ‘moron’. This was a composite image of two different photos, edited to make it look like certain letters of the logo were blocked. The picture was seemingly intended to be a joke, but it had also been shared alongside captions which indicated that many people believed it was real.
21. Real people were mistaken for AI-generated fabrications. Full Fact debunked one such claim involving Reform UK candidate Mark Matlock, whose image on party leaflets led social media users to speculate he might not be a real person, simply because he "looked AI-generated".
22. We also found parties leveraging paid display advertising to rapidly disseminate unfounded assertions about opposition policies to a targeted audience. Days before the election, the Conservatives published widely circulated online advertisements claiming that Labour’s plan to implement a “national ULEZ” (Ultra-Low Emission Zone) would be “coming to a road near you this July”. A search on Facebook’s Ad Library at the time suggested that more than 800 versions of the advert may have been posted. But there was no specific evidence that Labour was planning to introduce such a scheme, and the party denied any plans to do so. There were also no plans in Labour’s manifesto for a ‘national ULEZ’, and Full Fact was unable to find any other specific information to back up the claim.
23. Recent analysis has also shown how community notes systems can be manipulated by a relatively small network. In April 2026, an investigation by Indicator found that during the 2024 general election, five users on X worked together to remove community notes from accounts linked to the Conservative party. This reportedly led to a reduction in visible notes on tweets, including by Rishi Sunak. Alexios Mantzarlis, a former fact checker who led the research, wrote, “This operation wasn't always incorrect in its arguments, nor was it successful in every effort to remove notes. But the campaign shows that there is a political appetite to suppress Community Notes.” It is particularly concerning that politically-motivated actors might be able to benefit from this vulnerability, given Meta uses the same source code for its own system, and recently confirmed it plans to eventually roll out Community Notes globally.

Impact of misinformation from the 2024 UK election

24. The rough-and-tumble of election politics is nothing new, nor is it surprising that only four parties signed up to a Full Fact pledge calling for honest campaigning during the last election (Labour, the Conservatives and the Liberal Democrats did not).
25. While it is unlikely that a single piece of content – whether cheapfake or deepfake – would end up swaying significant numbers of voters in an election, there are other ways that misinformation and disinformation damage our democratic system and deepen the already low levels of trust in politics.
26. Confidence in elections and participation: According to an Electoral Commission survey after the 2024 general election, 61% of respondents said they saw misleading or inaccurate information about political parties’ policies during the campaign, and 52% said they saw misleading or inaccurate information about candidates.
27. Voter confusion: In 2024, Ofcom found that 60% of respondents saw false or misleading material about an election in a single week. Over half were concerned about the impact of deepfakes – but 46% were not sure if they had seen one. The growing volume of misinformation and realism of AI-generated content makes it harder for voters to identify trusted sources. CETaS’ Sam Stockwell describes voter confusion as a “second-order” category of damage: “UK users being confused over whether election content they viewed was synthetic or genuine—even on deepfakes which had been verified as such." Related to this, when there is widespread expectation of seeing deepfakes, and detection is challenging, this creates an opportunity for politicians and others to dismiss allegations that may turn out to be credible.
28. Erosion of trust in individuals and institutions: Deepfakes do not just contain misleading claims; they undermine trust more broadly by convincingly imitating trusted individuals and organisations. For example, a Full Fact investigation found that deepfake videos of academics and health leaders have been used to promote supplements. CETaS has highlighted a series of deepfakes of political figures in recent years – and found that these techniques do not just add perceived credibility to deceptive content, but damage public trust in the organisation or individual whose likeness has been used. Over time, this contributes to general scepticism in institutions, genuine sources and communications.

Concerns among the public and MPs since the election

29. As before the 2024 election, there is still widespread concern among politicians and the public that UK elections and candidates will be targeted by disinformation, whether homegrown or from overseas actors. 
30. An Electoral Commission survey in 2024 found that misinformation and disinformation are the second biggest concern in elections (70%), after media bias. Despite this, the survey found that three quarters (76%) of people do not think enough is being done to tackle mis- and disinformation in elections, and only 5% think sufficient action is being taken.
31. Full Fact recently commissioned polling by YouGov, which will be examined in our report on misinformation and democracy to be published in June 2026. It underlines the extent to which the public perceives inaction: two thirds of people (66%) think the government is doing too little to address AI-generated misinformation. Widespread public concern about misinformation corresponds with concern that not enough is being done to address the issue. This takes place against the backdrop of rock bottom public trust in our politics.
32. The government’s policy paper for the Representation of the People Bill, Restoring trust in our democracy: Our strategy for modern and secure elections, published in July 2025, recognised that “Our own democracy is being threatened by misinformation” and committed to restoring trust in politics. Days later, Sir Keir Starmer told the Liaison Committee that he was “very worried at the last election about misinformation, and I am very worried about the potential for misinformation in future elections in this country”.
33. When the Representation of the People Bill was first debated in Parliament in March 2026, numerous MPs on a cross-party basis called for the government to do more to tackle misinformation and disinformation. Those interventions are considered below. However, despite significant public and political concern about misinformation and disinformation in our democracy, the Bill includes limited measures that would address the issue.
34. Recent Parliamentary activity has shone a light on foreign influence risks. In launching the inquiry, Disinformation diplomacy: how malign actors are seeking to undermine democracy in 2025, the chair of the House of Commons Foreign Affairs Committee argued that disinformation campaigns are designed deliberately to sow the seeds of discontent. “They have been weaponised to subvert free and fair elections, to undermine the rules-based international order and to propagate anti-Western narratives. Foreign malign actors have realised the power of the media and social media in supporting their aims and interests.”
35. During a March 2026 evidence session of the Foreign Affairs Committee, the Chair asked representatives of Meta, X and TikTok about foreign inference in Romania and Moldova, where dormant accounts became active overnight during elections: “if they can do it in Moldova, they can do it in the UK.” Sir John Whittingdale challenged the companies’ willingness to address concerns about disinformation expressed to the Committee by NGOs and government institutions in France, Moldova, Estonia and the UK: “all of them have suggested to me that the platforms have not been wholly co-operative in trying to prevent it.”
36. During the second reading of the Representation of the People Bill in March 2026, Martin Wrigley argued that misinformation, disinformation and modern technology are “distorting our electoral position, while Alex Barros-Curtis said “we must urgently deal with disinformation and online operations” and noted that the Electoral Commission, Ofcom and the police “need the resources to deal with the threat of personalised algorithmic feeds and AI-enabled manipulation that feeds misinformation about our elections.” Rushanara Ali said that some MPs have wondered whether “if we had known what we know now about the state of harassment and intimidation in our politics, we would have stood for Parliament.”
37. Full Fact agrees with the Minister’s recognition during this debate that the issue of misinformation and disinformation “needs to be addressed more forcefully”. But rather than treating misinformation and disinformation as distinct policy issues which should be in or out of scope of online safety legislation or electoral law, MPs need to recognise that these types of information disorder use and are exacerbated by the same systems.
38. This is illustrated by Meta’s evidence to the Foreign Affairs Committee in 2026, where a spokesperson stated that of 250 co-ordinated inauthentic behaviour networks taken down by Meta across 70 countries, “Roughly half of those were foreign influence operations, so they originated in one country and targeted another. The other half were domestic or a mix of domestic and foreign, so they were targeting audiences in the country that they were based in.” Similarly, when announcing the Disinformation Diplomacy inquiry, Dame Emily Thornberry emphasised that threats aren’t just coming from hostile states, but also from non-state actors who have significant influence over our information environment. “Powerful figures such as Elon Musk,” she said, “exploit their platform to spread disinformation that disrupts and destabilises.”
39. The recent review into countering foreign financial influence and interference in UK politics, by Philip Rycroft, expressed concern about how the government counters hostile state online interference: “Responsibilities are dispersed across different departments and agencies with no apparent focal point. Government work is still divided between the foreign and domestic spheres, a distinction that is almost completely irrelevant in dealing with this problem.” Mr Rycroft called for various measures that would help improve the information environment for elections and safeguard UK democracy, including high-level responsibility to sustain our democracy and coordinate the response to threats.
40. Writing in Full Fact’s 2025 policy report, Sam Stockwell of the Centre for Emerging Technology and Security outlined the multi-faceted nature of misinformation during the 2024 election, arguing that the diversity of systems, motivations and actors involved in promoting misinformation “underscore[s] the need to tackle misinformation and disinformation more systematically, as opposed to just narrow election- or AI-based interventions.”

No foreign interference in the UK election was the exception, not the rule

41. The lack of known and publicised attempts at foreign interference in the UK’s 2024 general election may have been the exception rather than the rule. Other countries saw far more concerted efforts to influence voters.
42. A network of approximately 100 fake news websites—some of them set up years in advance—was activated by a Russian influence operation ahead of Germany’s federal elections in February 2025. False claims about a number of German politicians were created using AI and deepfake technology, including accusations of physical abuse and espionage. Meanwhile, Elon Musk sided openly with the far-right populist party, Alternative for Germany, regularly spreading false and misleading claims on X, in posts that were boosted by the algorithm he owns and which received millions of views.
43. Following investigative reporting by Ziarul de Garda (ZdG) during the Moldovan elections in October and November 2024, Moldovan police found that 300,000 people in Moldova had a Russian banking app on their phones that allowed them to receive money for helping to support a pro-Russian agenda. Other media outlets reported on similar vote-buying schemes. ZdG tried to talk to both Telegram and TikTok about their roles in hosting and disseminating disinformation but neither platform was responsive.
44. Two days before the polls opened for the 2023 Slovakian parliamentary elections, faked audio went viral, purporting to be one of the party leaders in conversation with a journalist and claiming to have rigged the election.
45. In Ireland, an AI-generated video falsely showed Catherine Connolly withdrawing from the 2025 presidential race days before voting.
46. Even a small number of credible-looking items can generate uncertainty far beyond any corrections, and content does not have to be persuasive to be harmful.

How the existing regulatory framework can be improved

47. In its Public Communication Scan of the United Kingdom published in December 2023, the OECD highlighted “a noteworthy gap in the legislative and policy landscape [...] on mis- and disinformation in the context of elections”.
48. Despite misinformation being the most encountered type of harm online, repeated warnings from experts about the risks of disinformation from hostile actors, and the growing crisis of public trust, the UK’s laws do little to protect our democracy from the harms of misinformation and disinformation.
49. In March 2021, the then Prime Minister said the Online Safety Act (OSA) would tackle collective online harms, including threats to democracy. The previous government noted “that misinformation and disinformation surrounding elections are a risk to democracy and it is vital to address this issue.” But the OSA failed to address the harms that misinformation and disinformation can cause to democracy.

56. The National Security Act 2023 included a new ‘foreign interference’ offence but there are practical challenges to enforcement and attribution, and the law did not combat domestic political deepfakes. The Elections Act 2022 did not tackle misinformation and disinformation in politics or address the harms they cause to democracy, and amendments relating to disinformation were rejected.
57. The government’s policy paper for electoral reform included some measures to deal with misinformation and disinformation. However, two measures in the policy paper that would have helped to address misinformation in UK politics were not in the Bill: a new code of conduct to ensure campaigns are conducted with honesty and integrity; and a requirement for campaign material to clearly identify the affiliated party.
58. The policy paper also recognised the need to “guard newly enfranchised young people against the potential effects of online mis- and disinformation”. Yet the Bill contains no measures to strengthen political and media literacy, despite recent research showing that only half of 13–17-year-olds feel confident identifying false political information.
59. Full Fact is aware of two amendments to the Representation of the People Bill relating to online safety which have been rejected by the Public Bill Office. This is a consequence of a bill that has not been drafted to address misinformation and disinformation in elections, and whose scope is focused on other areas of democratic risk.
60. Internet Matters has identified sustained concerns from parents about children’s ability and confidence in spotting misinformation. To address the urgent need to improve media literacy provision alongside the extension of the franchise, Full Fact proposed and helped to draft an amendment placing a statutory duty on platforms to provide media literacy. This was rejected on the grounds that the Bill’s purpose does not relate to Ofcom’s duties and functions, nor potential new duties to regulate online services providers to support media literacy.
61. The Online Safety Act Network proposed and helped to draft an amendment requiring Ofcom to introduce a code of practice to address online abuse of candidates during election periods. This built upon a growing body of evidence pointing to many harms to electoral democracy originating on, and being exacerbated by, online platforms, and extensive investigation by the Speaker’s Conference which concluded that “the Government should consider the merits of mandating Ofcom to produce an elections code of practice for social media platforms”. However, advice from the Public Bill Office stated that the Bill relates only to elections and the work of the Electoral Commission and therefore cannot be used as a means to amend the OSA nor give new powers or responsibilities to Ofcom.
62. As the Online Safety Act Network notes in its evidence to this committee, this means that these amendments and the issues they address - including one which was the central focus of the Speaker’s Conference inquiry - cannot be debated.
63. In its current form, the Bill fails to tackle the serious threat of misinformation and disinformation to democracy. Full Fact recommends the following measures to strengthen the Bill and electoral integrity.

Recommendations

64. In the light of our evidence, we make the following recommendations:

●        Upgrade the Online Safety Act to safeguard democracy

The Online Safety Act (OSA) was watered down from its original ambition to protect UK democracy from the harms of misinformation. The Bill should amend the OSA to require major platforms and search engines to address any negative effects they pose to civic discourse, elections and public security. Online services should also be explicitly required to mitigate election-related offences, including false statements about candidates and undue influence.

See Appendix 2 for draft amendment, although it is likely this would be rejected as out of the scope of the Bill.

●        Stronger rules to deal with political deepfakes

The Bill contains no measures to tackle the growing risk of AI-generated political content. Campaigners should be required to clearly label any AI-generated or digitally manipulated political content, including details of the individual and party responsible. The government should clarify whether the offence of a false statement about a candidate applies to digitally manipulated and AI-generated content. The Electoral Commission has called for the offence to be updated and it is currently unclear whether synthetic media would fall within the scope of the offence.

Lisa Smart and Emily Darlington have tabled this as NC18 and NC10 respectively.

●        Establish a comprehensive public library of political adverts

Independent expert bodies, including the Electoral Commission, have called for comprehensive digital libraries of political adverts. This would enable journalists, researchers and the public to scrutinise campaigns, hold parties and platforms to account, and identify misinformation and disinformation. The Bill should establish a comprehensive, publicly accessible and searchable library of all political adverts. Platforms hosting political advertising in the UK should be required to provide data on who paid, how much they spent, the intended audience, and targeting methods. Records should be preserved in the National Archives.

Emily Darlington has tabled this as NC24.

●        Regulate to prevent misinformation and disinformation in political ads

The honesty and accuracy of UK non-broadcast political advertising is not regulated. The Advertising Standards Authority and a parliamentary committee previously called for a regulatory committee to oversee political ads. The proposed code of conduct for campaigning should be placed on a statutory footing, with minimum standards on honesty, accuracy and transparency in political ads, including use of AI-generated content. Compliance should be overseen by a regulatory committee, with powers and resources to act swiftly during election periods.

Justin Madders has tabled this as NC61, NC62, NC63 and NC64.

●        Create a transparent system for dealing with electoral information incidents

The UK does not have a public protocol to manage incidents that may undermine the fairness or integrity of an election. The lack of transparency around electoral security bodies also poses issues for trust in our politics. The Bill should establish a clear, independent public protocol for managing election incidents, modelled on Canada’s approach. Decisions to alert voters and others should be taken independently of ministers and according to predefined criteria. This should be supplemented by more public transparency about electoral security bodies.

Emily Darlington has tabled this as NC26.

●        Increase the investigative powers of the Electoral Commission

The Electoral Commission does not have the power to obtain documents and information outside of a formal investigation. This means they cannot compel online platforms to provide information in real-time during an election, unless they open an investigation. The Commission has previously expressed concern about this gap. The Bill strengthens enforcement functions and expands information-sharing powers. Failing to expand the Commission’s information-gathering powers risks undermining the value of those important reforms.

Emily Darlington has tabled this as NC25.

●        Give platforms a statutory duty to provide effective media literacy

Newly enfranchised voters must be equipped to navigate today’s information environment, as our recent joint briefing explained. The Bill should introduce a statutory duty on major online platforms to support effective media and political literacy, backed by a code of practice with clear standards and accountability.

An amendment to this effect was rejected by the Public Bill Office as out of scope (see Appendix 1).

April 2026

Appendices

Appendix 1: Draft amendment to the Representation of the People Bill to impose a statutory duty on large online services to support media literacy

The Bill’s long title includes making provision to extend the right to vote to 16 and 17 year olds and for connected purposes. The government’s Strategy for modern and secure elections last year said “19. It is important that we guard newly enfranchised young people against the potential effects of online mis- and disinformation.” The notes to editors in the government’s press release for the Bill, focused extensively on measures to improve political and media literacy, including through changes to the national curriculum. Full Fact drafted this as an insert to the Bill rather than an amendment to the OSA, as we understand that anything relating to the Online Safety Act will be out of scope. This amendment is intended to go in Part 1, linked to the extension of the right to vote.

Duty to promote media literacy

1. This section applies to a provider of—

(a)     a Category 1 service, or

(b)     a Category 2A search service,

within the meaning of the Online Safety Act 2023 (a “regulated service”).

2. A provider of a regulated service must operate the service using systems and processes, proportionate to the nature and scale of the service, designed to secure and improve a high level of media literacy among users.

3. The duty under subsection (2) is a duty to take proactive and systemic measures, including by the design and operation of the service, to—

(a)     improve users’ ability to access and critically evaluate information encountered on the service;

(b)     support users in understanding how information is created, selected, presented and disseminated; and

(c)     enable users to engage confidently, safely and responsibly with information.

4. In complying with this section, a provider must have regard to—

(a)     the desirability of promoting media literacy as a core capability for participation in society, including participation in democratic processes;

(b)     the need to support users of different ages, abilities and levels of capability; and

(c)     the importance of ensuring that measures are accessible, inclusive and effective.

5. In this section, “media literacy” means the ability of users to—

(a)     access and critically evaluate information encountered through regulated services;

(b)     assess the accuracy, reliability, provenance and authenticity of content;

(c)     understand how content is created, selected, ranked, recommended, generated and targeted; and

(d)     engage with information in an informed and responsible manner.

Scope of duty and code of practice

6. The duty under subsection (2) requires providers to implement and operate systems and processes which may include the following—

(a)     functionalities that support users in assessing the credibility, provenance and reliability of content;

(b)     the provision of clear, accessible and timely contextual information about content;

(c)     the identification or labelling of synthetic, manipulated or AI-generated content where reasonably practicable;

(d)     user controls and transparency tools relating to the operation of recommender systems and content curation;

(e)     prompts, friction or other interventions designed to encourage reflective engagement with content; and

(f)      signposting to authoritative sources of information.

7. Providers must ensure that those measures are—

(a)     embedded in the design and operation of the service;

(b)     appropriate having regard to the nature and use of the service;

(c)     informed by evidence of effectiveness, including research, testing or user feedback where available; and

(d)     regularly monitored, evaluated and improved.

8. Ofcom must prepare and issue one or more codes of practice for the purpose of assisting providers of regulated services to comply with their duties under this section.

9. A code of practice under subsection (8) must—

(a)     set out minimum standards for compliance with the duties in this section;

(b)     include recommended measures relating to service design, transparency, user understanding and informed engagement;

(c)     include provision for assessing the effectiveness of measures taken, including the regular publication of accessible data on the impact of those measures; and

(d)     take account of the differing functionalities, design and capacities of regulated services.

10. In preparing a code of practice under subsection (8), Ofcom must—

(a)     consult the Secretary of State, Ofcom’s Online Information Advisory Committee and such other persons as Ofcom considers appropriate; and

(b)     have regard to—
(i) the need to promote a coherent and effective approach to media literacy across the United Kingdom,
(ii) the importance of ensuring that measures are accessible, inclusive and evidence-based, and
(iii) its functions relating to media literacy under the Online Safety Act 2023.

11. Ofcom must publish any code of practice issued under this section and keep it under review.

12. A provider must have regard to any code of practice issued under this section when complying with its duties.

13. A provider may, in lieu of measures set out in a code of practice, take alternative measures if it can demonstrate that those measures achieve an equivalent or greater level of effectiveness.

14. In determining whether a provider has complied with its duties, Ofcom must take into account compliance with a code of practice and any evidence as to the effectiveness of alternative measures.

15. Ofcom may require a provider to supply such information as it reasonably requires for the purpose of assessing compliance with this section.

16. The Secretary of State may by regulations make provision about minimum requirements to be met by providers of regulated services in complying with their duties under this section.

Appendix 2: Draft amendment to the Representation of the People Bill to impose systemic risk duties on large regulated services

The Online Safety Act (OSA) was watered down from its original ambition to protect UK democracy from the harms of misinformation. The Science, Innovation and Technology Committee concluded last year the Act “cannot keep the UK public safe as it was not designed to tackle misinformation.” The Bill should amend the OSA to require major platforms and search engines to address any negative effects they pose to civic discourse, elections and public security. Online services should also be explicitly required to mitigate election-related offences, including false statements about candidates and undue influence.

The Online Safety Act 2023 is amended as follows.

After section 34 insert—

34A Duty to assess systemic risks

(1)     This section applies to a provider of—

(a)     a Category 1 service, or

(b)     a Category 2A search service.

(2)     A provider to whom this section applies must carry out a systemic risk assessment in relation to its service.

(3)     A systemic risk assessment is an assessment of any systemic risk arising from the design, functioning, operation or use of the service.

(4)     In this section “systemic risk” means a risk arising from the design, functioning, operation or use of a service which may reasonably be expected to have actual or foreseeable significant adverse effects in the United Kingdom on—

(a)     civic discourse,

(b)     the integrity of electoral processes,

(c)     public security,

(d)     public health, or

(e)     the protection of fundamental rights.

(5)     The Secretary of State may by regulations further specify descriptions of the systemic risks referred to in subsection (4).

(6)     A statutory instrument containing regulations under subsection (5) may not be made unless a draft of the instrument has been laid before and approved by a resolution of each House of Parliament.

(7)     In carrying out a systemic risk assessment the provider must consider risks arising from—

(a)     the design, functioning or use of algorithmic systems, including recommender systems,

(b)     advertising and other monetisation systems,

(c)     the design and operation of features affecting the dissemination, virality or amplification of content,

(d)     the terms of service and governance systems of the service,

(e)     the manipulation of the service, including coordinated or inauthentic behaviour,

(f)      the collection, processing or use of personal data for targeting or recommendation purposes, and

(g)     content moderation systems, including automated detection and enforcement systems.

(8)     A provider must carry out a systemic risk assessment at least once every year and in any event prior to deploying functionalities that are reasonably likely to materially affect the systemic risks referred to in subsection (4).

(9)     Ofcom may by notice require a provider to carry out a systemic risk assessment where Ofcom reasonably considers that—

(a)     there is a significant change to the design, functioning, operation or use of a service, or

(b)     circumstances give rise to a heightened systemic risk.

(10) A provider must periodically evaluate the effectiveness of measures taken under section 34B and update those measures where necessary to address systemic risks identified in subsequent assessments.

(11) A provider must—

(a)     keep a written record of each systemic risk assessment and its findings,

(b)     provide a copy of the record to Ofcom as soon as reasonably practicable after the assessment is completed, and

(c)     provide such additional information relating to the assessment as Ofcom may reasonably require.

34B Duty to mitigate systemic risks

(1)     A provider to whom section 34A applies must take proportionate measures relating to the design or operation of the service to mitigate and manage systemic risks identified in a systemic risk assessment.

(2)     Measures taken under this section may include measures relating to—

(a)     the design or operation of a recommender system or ranking system,

(b)     the design or operation of an advertising or monetisation system,

(c)     the detection and disruption of coordinated manipulation of the service, including coordinated inauthentic behaviour,

(d)     the design of a feature affecting the virality or amplification of content,

(e)     the availability and visibility of reliable information from public authorities or other trusted sources identified by Ofcom during an election, public emergency or other period of heightened risk, and

(f)      limitations on the targeting or amplification of content during periods of heightened systemic risk.

(3)     Measures taken under this section must—

(a)     be proportionate to the nature, severity, likelihood and reasonably foreseeable impact of the systemic risk identified, and

(b)     be implemented with due regard to the importance of protecting users’ rights to freedom of expression and access to information.

(4)     Compliance with this section concerns the systems and processes used in the design and operation of the service, and does not require a provider to determine the legality or accuracy of individual items of content except where required elsewhere in this Act.

34C Codes of practice: systemic risks

(1)     Ofcom must prepare and issue codes of practice giving guidance about compliance with sections 34A and 34B.

(2)     In preparing a code of practice under this section Ofcom must consult—

(a)     the Electoral Commission,

(b)     public health authorities,

(c)     independent researchers and civil society organisations with expertise in misinformation, disinformation, information integrity and online harms, and

(d)     such other persons as Ofcom considers appropriate.

(3)     A code of practice under this section may in particular include guidance about—

(a)     methodologies for systemic risk assessments,

(b)     appropriate mitigation measures relating to the design or operation of services,

(c)     record-keeping and transparency requirements,

(d)     engagement with public health and electoral authorities, and

(e)     safeguards to ensure that mitigation measures do not unduly restrict lawful speech.

Consequential amendments

In section 85 (codes of practice) of the Online Safety Act 2023, after subsection (1) insert—

(1A) Ofcom must prepare codes of practice under section 34C (codes of practice relating to systemic risks).

In Schedule 8 to the Online Safety Act 2023, after subsection (36) insert—

(36A) Measures taken or used by a provider for the purpose of complying with sections 34A or 34B (systemic risk duties).

April 2026