# AI rules in United Kingdom

No AI bill (none in the May 2026 King's Speech); AI is governed by existing regulators plus targeted criminal laws on intimate deepfakes and nudification tools and reformed automated-decision rules under the Data (Use and Access) Act 2025.

- **Approach:** Sector regulators and guidelines
- **Key bodies:** DSIT, AI Security Institute, ICO, Ofcom, CMA, FCA, Intellectual Property Office
- **Strategy:** [AI Opportunities Action Plan](https://www.gov.uk/government/publications/ai-opportunities-action-plan) (2025)
- **Last reviewed:** 4 Oct 2026

## Rules (7)
- [AI Opportunities Action Plan](https://wheresthe.ai/rule/gb-ai-opportunities-action-plan-2025/index.md): In force, Strategy, effective 13 Jan 2025. Government plan (with 50 recommendations accepted) to expand compute, data access, AI adoption in the public sector and regulator-led sandboxes, within the UK's non-statutory, regulator-by-regulator approach to AI. A one-year-on progress report was published in January 2026.
- [DUAA automated decision-making reforms](https://wheresthe.ai/rule/gb-data-use-and-access-act-2025-automated-decisions/index.md): In force, Law, effective 5 Feb 2026. Replaces UK GDPR Article 22 with new Articles 22A-22D: solely automated decisions with legal or similarly significant effects are broadly permitted on any lawful basis if safeguards apply (information, ability to contest, human intervention), while stricter limits remain for special-category data. Applies to all controllers using automated decision-making, including AI.
- [Deepfake intimate image creation offence](https://wheresthe.ai/rule/gb-duaa-2025-s138-purported-intimate-images/index.md): In force, Law, effective 6 Feb 2026. Inserts sections 66E-66H into the Sexual Offences Act 2003, making it an offence in England and Wales to create, or ask someone to create, a purported (including AI-generated) intimate image of an adult without consent or reasonable belief in consent. Applies to individuals.
- [OSA intimate image (incl. deepfake) sharing offences](https://wheresthe.ai/rule/gb-online-safety-act-2023-intimate-image-offences/index.md): In force, Law, effective 31 Jan 2024. Makes it an offence to share, or threaten to share, a photograph or film that 'shows, or appears to show' another person in an intimate state without consent, which captures AI-generated deepfakes. Platforms in scope of the Online Safety Act must treat such content as illegal content under Ofcom's codes.
- [Crime and Policing Act 2026 (nudification tools, AI CSAM generators)](https://wheresthe.ai/rule/gb-crime-and-policing-act-2026-ai-provisions/index.md): Partly in force, Law, effective 29 Apr 2026. Creates an offence of making, adapting or supplying a tool for generating purported intimate images (nudification apps), with a defence for taking all reasonable steps to prevent non-consensual use; creates offences relating to AI 'child sexual abuse image-generators'; and gives the Secretary of State power to extend Online Safety Act duties to providers of AI services regarding illegal AI-generated content. Applies to developers, suppliers and online services.
- [UK Copyright and AI report](https://wheresthe.ai/rule/gb-copyright-ai-report-2026/index.md): In force, Advisory. Statutory report following the Dec 2024-Feb 2025 consultation: the Government states that a broad text-and-data-mining exception with rights-holder opt-out is no longer its preferred way forward and that it will not change copyright law until confident reforms meet its objectives, instead gathering more evidence on transparency, licensing, technical standards and enforcement. Relevant to AI developers and rights holders.
- [Regulating for Growth Bill (sandboxes incl. AI)](https://wheresthe.ai/rule/gb-regulating-for-growth-bill-2026/index.md): Proposed, Bill. Announced bill to create cross-economy regulatory sandbox powers allowing controlled live-market trials in which existing legal requirements can be modified or suspended for technologies including AI, building on the planned 'AI Growth Lab'. It would apply to firms admitted to sandboxes.

## Programmes and facts
- AI Safety Institute renamed AI Security Institute on 14 Feb 2025
- UK coordinates the International Network for Advanced AI Measurement, Evaluation and Science (since Dec 2025)
- Copyright and AI report (18 Mar 2026): opt-out TDM exception no longer the preferred option

---
Canonical page: https://wheresthe.ai/j/gb/ · Not legal advice.