44 No. 78 JUSTICE rights, consumer safety, and intellectual property. It would also indirectly address AI-generated disinformation and bias through oversight mechanisms that could halt pilots involving harmful outputs like deepfakes eroding public trust or biased algorithms reproducing stereotypes. However, it avoids prescriptive measures like mandatory content labeling or bans, prioritizing evidence-gathering from tests to refine future policies and maintain public confidence in the benefits of AI. This approach is not reassuring when it comes to combating AI-generated antisemitism online. Under the current framework, no regulator has explicit, comprehensive responsibility for AI-generated hate including anti-Jewish racism. Ofcom’s remit extends only to content hosted on regulated services, not to the design or deployment of generative AI systems themselves. This means that AI tools capable of producing antisemitic text, images, or videos may fall largely outside regulatory scrutiny unless their outputs are disseminated on covered platforms. Legal and policy analyses by Deloitte,29 White & Case,30 and Kennedys Law31 have all noted that this creates a regulatory gap. While developers may adopt voluntary safeguards, there is no binding obligation to assess or mitigate the risk that AI systems will be used to generate racist or antisemitic content. Crafting a Balanced Regulatory Framework for AI in the United Kingdom to Combat Antisemitism As artificial intelligence technologies advance, they pose risks of amplifying societal harms such as antisemitism through generative tools that produce deepfakes, disinformation, and biased content. In the UK, existing frameworks like the OSA address platform dissemination of harmful material but fall short in regulating AI creation at the source. This article proposes targeted regulatory changes to mitigate AI-generated antisemitism, emphasizing transparency, accountability, and enhanced moderation. These measures are essential to prevent the unchecked spread of manipulative content that blurs reality and incites hate, while preserving innovation and free speech through proportionate, principles-based implementation. The discussion concludes with considerations on whether these reforms should amend the OSA or form a standalone AI bill. Suggested Approach for Regulating Harmful AI Content Several regulatory adjustments are required to address the unique challenges of AI-generated antisemitism, where tools can rapidly create realistic but fabricated content that perpetuates stereotypes and disinformation. First, mandating transparency in AI outputs, such as requiring watermarks or metadata to label generated material, is crucial. This is necessary because current systems often produce indistinguishable deepfakes, eroding public trust and enabling emotional manipulation that fuels hate. By making origins clear, users can critically evaluate content, reducing the likelihood of sharing deceptive narratives without infringing on free speech, as labeling informs rather than censors. This aligns with the government’s 2023 pro-innovation AI principles, which already emphasize transparency to manage risks like misinformation. Second, generative AI models should undergo mandatory pre-deployment risk assessments for bias and hate amplification, including audits of training data to identify and mitigate discriminatory patterns. Such assessments are needed because untrained or poorly curated data can reproduce societal biases, leading to outputs that subtly embed harmful tropes. Implementing these would make a difference by preventing harmful content at the development stage, thus decreasing its proliferation on platforms and lowering the burden on downstream moderation. Third, the establishment of a dedicated AI Authority, as proposed in ongoing legislative discussions, would provide centralized oversight, monitoring compliance with principles and issuing sector-specific guidance on risks like 29. Valeria Gallo Suchitra Nair, “The UK’s framework for AI regulation” (Feb. 21, 2024), available at https://www. deloitte.com/uk/en/blogs/ecrs/the-uks-framework-for-ai-regulation.html 30. Daniel Mair, Jeffrey Shin and Zarlush Zaidi, “AI Watch: Global regulatory tracker – United Kingdom” (Nov. 25, 2025), available at https://www.whitecase.com/insight-our-thinking/ai-watch-global-regulatory-tracker-united-kingdom 31. Dr. Nathalie Moreno, “The Artificial Intelligence (Regulation) Bill: Closing the UK’s AI Regulation Gap?” (March 7, 2025), available at https://www.kennedyslaw.com/en/thought-leadership/article/2025/the-artificial-intelligenceregulation-bill-closing-the-uks-ai-regulation-gap/
RkJQdWJsaXNoZXIy MjgzNzA=