In this article Taylor Hampton Solicitors examines the growing interest and concerns surrounding copyright and artificial intelligence in light of a recent Government consultation.
On 13 January 2025, the UK Prime Minister Sir Keir Starmer announced an Action Plan to help the UK become an “AI superpower”. Then, on 18 March 2026, following a public consultation, the UK Government published a report on Copyright and Artificial Intelligence. The report looked at the inherent tension between ensuring that there is sufficient protection for copyrighted work, whilst also encouraging AI development. The public consultation received over 11,000 responses from a range of industries, including academics, legal professionals, businesses and representative groups, reflecting the strength of cross-sector feeling. The report considered a number of different proposals and options moving forward, with a view to shaping future Government policy.
Although the report is detailed in content, the conclusion is broadly one of maintaining the status quo, to allow the Government further time to gather evidence and consider the best way forward for all impacted.
Background
The ONS’s Business Insights and impact on the UK economy showed that in December 2025, 25% of UK firms had adopted some form of AI. This is up from 15% in 2023, showcasing the speed at which AI is being adopted within businesses and industry sectors.
However, the growth of AI presents risks, particularly to the creative sectors, as AI models need to train on access to billions, if not trillions, of datapoints, and typically exploit publicly available material to do so. If this training is carried out without sufficient consent or renumeration to the original authors or owners of that material, then it may give rise to a claim for copyright infringement. This conflict makes promoting AI, whilst protecting original works, a difficult balancing act for the Government.
The divide in feeling between creatives and AI developers is apparent. For example, according to the Independent Society of Musicians 2026 report, 99% of creators believe their work has been scraped (i.e. used to train an AI model dataset) without their consent. On the other hand, OpenAI has stated that it would be “impossible to train today’s leading AI models without using copyrighted materials”.
Current Copyright Law
The Copyright, Designs and Patents Act 1988 (‘CDPA’) sets out the relevant framework for copyright protection. Copyright does not need to be registered; it either exists or it does not exist. This means that a claimant in an action for copyright infringement will, in the first instance, need to prove:
- Copyright subsists; and
- They are the owner or author of that copyright.
Generally, copyrighted material can only be used with consent from its owner. This is typically by way of a licence or assignment. However, in relation to AI models, it can be very difficult for a copyright owner to ascertain whether their work has been exploited. This is because the public typically only has access to the output of an AI system, rather than insight into whatever data has been inputted for training purposes.
There is, in theory, an exception to the requirement for consent. This is under s. 29A CDPA which permits the copying of information without consent where it is used “for the sole purpose of research for a non-commercial purpose”. This is also known as the ‘text and data mining exception’. It is unlikely that such an exception, as it currently stands, will apply to commercial entities such as Open AI or Anthropic.
Findings and Conclusion
The report found that AI developers tended to favour the introduction of a much broader text and data mining exception, which allows for data mining on copyrighted works for any purpose, with no ability to opt out. On the other hand, creatives tended to favour strengthening existing copyright principles so that there is an obligation on AI companies to enter into licences in order to use copyrighted works for AI development.
The Government, faced with this impasse, has decided for now not to intervene directly in the licensing market. It considers that (at this stage) proposals to review current transparency rules relating to training material used in AI models, which would allow a copyright owner to more easily establish if their work has been used for AI training and therefore assert their rights more effectively, may be sufficient alongside current market-led licencing.
It is clear that the status quo satisfies neither side. With both sectors too economically significant to leave in a state of legal uncertainty indefinitely, the Government may need to move beyond observation if it is serious about the country’s AI ambitions.
Contact Us
If you suspect that your copyright or other intellectual property rights have been infringed, contact us here for a free consultation. The team at Taylor Hampton are specialists in the field and can expertly assist with enforcing your rights. Visit our services page here for more information.