Written evidence submitted by Dr Elena Abrusci (SMH0050)

House of Commons Science, Innovation and Technology Committee Inquiry: Social media, misinformation and harmful algorithms

This submission is prepared by Dr Elena Abrusci, Senior Lecturer in Law at Brunel University of London. Dr Abrusci is a renowned expert in human rights and technology, with a specific research focus on the impact of emerging technologies on human rights and digital regulation. She’s a member of the Brunel Centre for Artificial Intelligence: Social and Digital Innovation and of the Global Academic Network of the Centre for AI and Digital Policy (CAIDP). Dr Abrusci served as a Policy Advisor on Digital Regulation with the Department of Digital, Culture, Media and Sport in 2020-21 and regularly support UK governmental departments, Parliamentary Committees and EU institutions in AI-related policy development.

This submission mainly addresses the following questions of the Terms of Reference:

• How effective is the UK's regulatory and legislative framework on tackling these issues?
• How effective will the Online Safety Act be in combatting harmful social media content?
• What more should be done to combat potentially harmful social media and AI content?

1. Misinformation and disinformation are not new phenomena in society, yet the policy answers have been limited and overall ineffective. The challenges posed by online forms of misinformation and disinformation have been known to policy makers for more than a decade now, spanning sectors and contexts. Both governments and social media platforms around the world have put in place a wide range of countermeasures, ranging from anti-disinformation laws, duty of care, content moderation and curation policies and media literacy programmes. Nevertheless, these interventions have had limited impact in addressing and reducing the harm to individuals and society.
2. The development and spread of generative AI tools in recent years has enabled more actors to engage with malicious content and further increased the amount of online misinformation and disinformation. However, generative AI has not changed the nature of the problem, nor the harm it causes to individuals and society.  Likewise, the responses to misinformation and disinformation produced through generative AI tools should be the same for any type of online misinformation and disinformation.
3. Although its promising ambition, the UK Online Safety Act is not adequate and to address the online misinformation and disinformation phenomenon. This is because it does not strike a good balance between freedom of expression and the protection of the rights of others, contains unclear and excessively broad definitions, and it does not provide relevant regulators (not only Ofcom but also the Equality and Human Rights Commission, the Advertising and Standards Authority, and the Electoral Commissions among others) with the needed powers and remit to enforce it and apply it to all the different contexts and situations where online misinformation and disinformation can spread and cause harm.
4. Any intervention aimed at regulating the public debate needs to carefully balance the freedom of expression of who wants to speak and the rights of those who may be impacted, directly or indirectly, from that content.  This difficult assessment is a legal obligation the United Kingdom has under its 1998 Human Rights Act (Article 10) and under its international obligations stemming from the European Convention on Human Rights (Article 10) and the International Covenant on Civil and Political Rights (Article 19). These legal provisions requires states to follow a three-parts test and allow only restrictions that are i) prescribed by law; ii) pursuing a legitimate aim and iii) necessary in a democratic society. Failure to do so, will result in a violation of freedom of expression.[1]
5. The Online Safety Act does not strike a fair balance between these competing interests because it imposes confusing duties on service providers that are likely to either be implemented through harsh content moderation policy or be completely disregarded in an ideological fight for free speech. Although it is important to intervene to regulate the use of algorithms that social media platforms employs to prioritize harmful content on users’ feeds just to increase their advertising profit, intermediary liability for users’ content should be carefully designed.[2] Definitions should be extremely clear and rooted in international human rights law and regulators should play a bigger and more significant role in monitoring and overseeing the case-by-case implementation of these obligations. 
6. The Online Safety Act contains a new false communication offence that applies to content that could cause ‘non-trivial psychological or physical harm’ to users. While this shows a clear commitment to address the issue of online misinformation and disinformation, the definition leaves a too wide margin of interpretation to the service providers as to what content meets the required threshold.[3] It therefore delegates to private actors, which are not under public scrutiny, the final decision of what content should be kept online and what should be removed. This inevitably leads to either over-restriction of freedom of expression or lower protection of the rights of those affected and targeted by the harmful content.
7. The Online Safety Act does not mention at all the phenomenon of deepfakes, which are at the heart of some of the most intrusive and effective forms of misleading information. AI-generated or altered videos and pictures have the potential of being extremely effective in influencing users’ thoughts and behaviours, but the Act does not acknowledge their existence to the point that this could become a legal loophole for future implementation. Ofcom has recently conducted important research on this matter,[4] but it also recognises the limits of the Act in banning all types of deepfakes.
8. The Online Safety Act entrusts Ofcom with the key role of enforcing the provisions contained therein and much of the implementation and details is left to secondary legislation that is still to be approved. The role of regulators in addressing misinformation and disinformation is paramount to fill the gaps of the text of the Online Safety Act and ensure that broad and general definitions and provisions and implemented in a way that respects human rights and protect from harms. However, the Act could be clearer on the exact powers Ofcom has and should further ensure that Ofcom is adequately resourced to carry out the enormous task of monitoring the whole online space.
9. As misinformation and disinformation take place in different contexts, through different channels and have a wide harmful impact, its regulation and monitoring should not be left only to one single regulator. In addition to Ofcom, there is the need for providing other regulators with the powers, remits and resources to effectively intervene in this field. Regulators such as the Electoral Commission (for election-related disinformation), the Advertising and Standards Authority (as mis/disinformation is deeply linked to online advertising) and the Equality and Human Rights Commission (for the impact misinformation and disinformation have on a wide range of human rights) should all be included in the implementation and enforcement of the Online Safety Act.
10. This multi-regulator effort should also be carried out in partnership with the Digital Regulation Cooperation Forum, which should not leave the matter of misinformation and disinformation in the hands of just one regulator. On the contrary, the challenge of tackling the harm of misinformation and disinformation could be an opportunity for the DRCF to extend its membership to other regulators relevant in addressing digital matters.

18 December 2024