UK to consult on deepfake protections this summer
Key Points
- UK government will consult on legal protections against unauthorised AI deepfakes in summer 2026
- AI labelling taskforce to publish interim report in autumn 2026 on marking AI-generated content
- Ministers no longer favour broad copyright exception with opt-out mechanism for AI training
- Sovereign AI Unit launched April 2026 with £500 million funding to support British AI firms
- Government response published 15 May 2026 alongside chair's call for mandatory transparency rules
The UK government will consult this summer on new legal protections for people whose image or voice is copied by AI without their consent.
The commitment was confirmed in the government’s response to the House of Lords Communications and Digital Committee report on AI, copyright and the creative industries, published on Friday (15 May).
The consultation will seek views on how to address harms caused by unauthorised digital replicas, while protecting legitimate uses of the technology. It forms one of four areas the government has identified for action in the coming months.
Existing UK law leaves victims of AI deepfakes with limited routes to redress. Members of the public cannot rely on the tort of passing off, which requires evidence of commercial goodwill, while copyright law only applies where a substantial part of an existing work has been copied.
Performers’ rights are tied to recordings of existing performances and offer no protection where an AI generates a new performance based on someone’s voice or likeness.
The government said it will monitor developments in other countries, including work by the US Copyright Office and the Danish government on similar protections.
A separate taskforce on AI labelling will publish an interim report in the autumn, with proposals on best practice for marking AI-generated content. “It can be helpful to consumers to understand whether content has been made using AI,” the government said in its response.
Labelling may also help protect against disinformation and harmful deepfakes, it added.
The taskforce will run alongside a review of technical mechanisms available for creators to control how their works are used online.
AI and copyright
The wider response confirms that ministers no longer have a preferred option on copyright reform for AI training, after the original proposal for a broad exception with an opt-out mechanism was rejected by most consultation respondents.
Illegal AI-generated content is already covered by the Online Safety Act, which requires in-scope services to assess and mitigate risks from illegal deepfakes.
The False Communications Offence makes it illegal to knowingly spread false information with the intent to cause non-trivial psychological or physical harm.
In April, the government launched a Sovereign AI Unit backed by £500 million across investment and research funding, designed to support British AI companies in priority domains and reduce reliance on overseas providers.
Updated Relationships, Sex and Health Education guidance taking effect from September 2026 will cover AI, deepfakes and online misogyny in schools.
The Curriculum and Assessment Review has also committed to strengthening media literacy in the updated national curriculum, including how to identify AI-generated content.
Baroness Keeley, who chairs the Communications and Digital Committee, criticised the government’s reliance on voluntary best practice rather than statutory transparency rules for AI developers.
She said only a mandatory framework would create the level playing field needed to encourage responsible training data practices among AI firms.
“Best practice alone will not promote licensing, drive compliance or enable robust enforcement,” said Keeley.
