Copyright and Generative AI

On 19 June 2025 the Data (Use and Access) Act (the "DUA Act") received Royal Assent and became law in the UK, having been passed by the UK Parliament on 11 June 2025. The DUA Act principally reforms the General Data Protection Regulation in the UK (the "UK GDPR") and the Privacy and Electronic Communications Regulations 2003 ("PECR"). This article focuses on the current position in the UK on copyright issues related to generative AI. See our articles on the changes the DUA Act makes to UK GDPR and PECR.

Before the DUA Act was passed by the UK Parliament, it was subject to an extended back and forth between the two houses of Parliament, in relation to the training of AI tools on copyright works. The House of Lords proposed an amended version of the DUA Act, which would include a requirement improving the position of copyright holders who are concerned that their works are being used to train AI models without their consent. 

This amendment was proposed to address the concerns of content creators that had been raised during a separate, contentious debate about the UK government's consultation on AI and copyright at the end of 2024, in which the government proposed to expand the existing general text and data mining exception to copyright law and place a new burden on copyright owners to pro-actively opt out if they do not want their works to be used to train generative AI models.  Many high profile content creators were vocal in their opposition to the proposals made in that consultation, which closed in February 2025.

The House of Lords' amendment to the DUA Act would have required the UK Government to implement regulations requiring organisations that operate web crawlers and general purpose AI models to publish transparency statements in which they identify the tools they are using to crawl the internet and how intellectual property owners can contact them regarding the use of their works.  This amendment could have made it easier for content creators to identify which works were being used to train AI models and potentially take action to preserve their rights.

The House of Commons rejected the amendment, arguing that the government is still considering the responses it received to its consultation on AI and copyright and legislation should not be passed before that has been completed. In addition, the UK government argued that the inclusion of transparency requirements for AI models in the DUA Act would overcomplicate the Act, and therefore requires separate legislative action. The government's concerns were in the context of ongoing litigation in the UK High Court, the outcome of which will have significant ramifications regarding the use of copyright materials in AI model training. The Court's ruling is expected later this year.

Ultimately, the UK government agreed within nine months of Royal Assent of the DUA Act to:

1. Publish a report on the use of copyright works in the development of AI systems.
2. Publish an economic assessment of the impact of each option on copyright owners and developers or users of AI systems described by the government in the Copyright and AI Consultation Paper.

The government must give a progress statement on both of these in six months.

Comment

The training of generative AI models on works protected by copyright and other intellectual property rights is a hot topic globally and there is ongoing litigation in the courts and debates related to regulatory reform in this area which may affect the ultimate position adopted on this issue in the UK and internationally.

The UK government's latest copyright and AI consultation aims to protect and strengthen both the creative and AI sector. It is expected that the AI copyright report, economic assessment and government proposals will be published in Spring 2026 and any new draft AI copyright laws could be passed by Parliament by late 2026.