Guide
AI and copyright: what businesses need to know
The current UK legal position on AI and copyright, including training data risks, AI-generated content ownership, and pending government reforms.
Artificial intelligence raises significant copyright questions for businesses. There is no AI-specific copyright law in the UK yet. The government is actively developing policy through consultation, industry working groups, and a statutory reporting obligation under the Data (Use and Access) Act 2025.
The key areas businesses need to understand are:
- training data rights — can you legally use copyrighted material to train AI models?
- AI-generated content ownership — who owns the output from generative AI tools?
- pending reforms — what changes are coming and when?
This guide explains the current position and what your business should do now.
Current legal position
The lack of AI-specific copyright legislation creates uncertainty for businesses. Existing copyright law under the Copyright, Designs and Patents Act 1988 (CDPA) was not designed with generative AI in mind. Courts have not yet tested how these rules apply to modern AI systems.
The government ran a consultation on AI and copyright from December 2024 to February 2025. The consultation attracted significant interest from both the creative industries and the technology sector. In December 2025, the government published a statement of progress confirming that two industry working groups had been established to develop practical solutions around transparency (how AI developers disclose what training data they use) and technical measures (how rights holders can control use of their content).
The DUAA 2025 (sections 135–137) requires the Secretary of State to produce a copyright report and economic impact assessment by 18 March 2026. This report will shape future legislation.
Training data risks
The practical implications for businesses are significant. If you use third-party AI tools such as large language models, image generators, or code assistants, you should check the provider’s terms of service regarding copyright compliance of their training data. Some providers have secured licences from rights holders. Others have not, and this creates downstream risk for your business.
If your business is training its own AI models, seek legal advice before using copyrighted material. The UK’s text and data mining exception under CDPA 1988 section 29A applies only to non-commercial research. There is no broad commercial exception. This is different from the position in some other jurisdictions.
Key risk areas include:
- Scraping web content to build training datasets without permission
- Fine-tuning models on proprietary or licensed content
- Using AI outputs that closely reproduce copyrighted source material
AI-generated content ownership
The CDPA 1988 section 9(3) provides for copyright in computer-generated works — works generated by a computer in circumstances where there is no human author. Copyright belongs to the person who made the arrangements necessary for the creation of the work.
This provision is unusual. Most jurisdictions require human authorship for copyright to exist. The United States, for example, has ruled that purely AI-generated works cannot be copyrighted.
For computer-generated works, copyright protection lasts 50 years from the date the work was made. This is shorter than the standard life-plus-70-years term for works with a human author.
However, the application of section 9(3) to modern generative AI is untested in court. If your business creates content using AI tools, you may own copyright under this provision, but there is no guarantee. Key uncertainties include:
- Who counts as the person who “made the arrangements” — the user, the AI developer, or both?
- Whether prompting an AI model constitutes sufficient “arrangements” for copyright to arise
- Whether works with significant human editing or curation have a human author (and therefore get the longer protection term)
Until courts or legislation clarify these questions, treat AI-generated content with caution. Do not rely on copyright protection for commercially valuable AI-generated works without legal advice.
What’s changing
Several developments will shape the future of AI and copyright in the UK:
- Copyright report (March 2026): DUAA sections 135–137 require the Secretary of State to report on AI and copyright, including an economic impact assessment. This is due by 18 March 2026 and will inform future legislation.
- Government AI Bill (second half of 2026): Expected to include provisions on AI and copyright. The scope and content will depend on the copyright report and working group outcomes.
- Industry working groups: Two groups are developing practical solutions — one on transparency (how AI developers disclose training data use) and one on technical measures (how rights holders can control access to their content).
- EU position: The EU has broader text and data mining exceptions under the Digital Single Market Directive (Articles 3–4). Rights holders can opt out of commercial TDM. UK businesses operating across both markets need to understand both regimes.
The direction of travel suggests some form of transparency obligation for AI developers is likely. Whether the UK introduces a broader commercial TDM exception or a licensing framework remains to be seen.