Summary
- The issues: AI developers need to use copyrighted works (extracted via data mining) to train AI models, but rightsholders argue that this infringes on their legal rights. A UK AI strategy was clearly needed.
- What has happened: in 2024 the UK Government proposed amending the law to facilitate the use of copyright protected material by AI developers in order to train AI models.
- The proposals: four options were considered, from stricter licensing rules to broader protections for AI developers.
- Response: the UK Government received a considerable amount of pushback due to competing concerns from rightsholders and AI developers.
- Further report: in response to this pushback, the UK Government has issued a further report into the proposed options in light of the feedback and the potential economic consequences
- Conclusions: No immediate legal changes will be made as trade-offs between AI innovation and creator protection remain unresolved.
- Further action: Focus will instead shift to different areas of the UK AI strategy consultations on deepfakes, AI labelling, creator protections, and licensing systems (such as the proposed “Creative Content Exchange”).
On 18 March 2026, the UK government published a report following a lengthy consultation into potential changes to UK IP legislation in response to growing concerns amid training generative AI models through mining data which is publicly accessible but subject to copyright or other intellectual property rights (a matter we have covered previously).
Whilst the consultation was aimed primarily at the creative industries, its conclusions, albeit frustratingly indecisive, have implications for generative AI developers and rightsholders across all sectors.
Context Behind the Consultation into AI and Copyright
Even since the early years of generative AI development, there has been an intense debate surrounding how rightsholders can expect to not have their intellectual property rights infringed by generative AI models trained through data mining.
This global issue has been particularly resonant in the creative industries; in the past few years alone, we have seen major conglomerates such as Universal Music Group bringing seismic infringement claims against generative AI developers, AI infringement being cited as a motivation behind one of the largest strikes in Hollywood history.
In response, generative AI developers have been claiming that data mining is necessary to develop cutting edge systems. From a US law perspective, the biggest developers in Silicon Valley have defended copyright infringement claims on the basis that data mining and AI training constitutes “fair use.” There is no equivalent legal defence in the UK. As such, the UK’s AI industry has been pushing for broader (or at least, clearer) usage rights in order to make the UK the European leader in AI development. Rightsholders, primarily those in the creative industries, are seeking tighter restrictions on data mining and the use of their material for AI training purposes.
To address these concerns, in December 2024 the UK Government consulted on reforms to the law of copyright in connection with the training of AI models.
Proposed Changes to IP Law and the UK AI strategy
The consultation explored four possible approaches to amending the law in light of the concerns of both the creative industries and AI developers.
- Option 1: no changes. The first option was to make no changes to the law. Under this model, the existing copyright framework would remain in place and AI developers would continue operating within current legal constraints.
- Option 2: strength copyright law. The second option involved strengthening copyright protections by requiring licences in all cases where copyrighted material is used for AI training. This includes the licencing of copyrighted works owned by UK rightsholders even though the AI training takes place abroad which is currently (in a very broad sense) permitted under existing intellectual property legislation.
- Option 3: text and data mining exception. The third option proposed introducing a broad text and data mining (TDM) exception to copyright law with no rights reservation mechanism and no ability for creators to opt out. This would effectively permit AI developers to use copyrighted material for training without obtaining permission.
- Option 4: text and data mining exception with opt-out clause. The fourth option proposed a middle-ground approach: a broad TDM exception combined with a rights reservation mechanism allowing creators to opt out of AI training uses.
Subsequent events
The consultation closed in February 2025. Shortly after in June 2025, the UK Government put forward the Data (Use and Access) Bill which proposed limited amendments to the Data Protection Act 2018 and the UK GDPR.
As part of the legislative process, significant concerns were raised by the UK’s House of Lords around the use of copyrighted works for AI training. Many members of the Lords wanted to use the Bill to protect the creative industries and cut short the UK Government’s consideration of responses to the consultation.
In order to get the Bill through Parliament, it was agreed that the UK Government would, within 9 months after the Data (Use and Access) Act 2025 became law, publish a report into the use of copyright protected work in the development of AI systems. The report effectively follows up on the outcome of the consultation.
The outcome of the report on the UK AI strategy
The consultation considered the responses to, and applied an economic impact assessment in respect of, the four options set out in the consultation.
The responses and the economic impact assessment, in the view of UK Government, revealed the difficulty of finding a legal solution that would be complimented with commercial success in both the AI and creative sectors.
- Option 1: no changes. The “no change” option was viewed as potentially restrictive for AI development. According to the assessment, existing UK copyright law could continue to act as a significant constraint on competitive general-purpose AI model training within the UK. In turn, this could inhibit broader AI development and adoption.
- Option 2: strengthening copyright law. Strengthening copyright protections by requiring licences in all cases, including in respect of AI models trained abroad, also raised concerns. The assessment suggested that overseas AI developers could potentially withdraw from the UK market entirely if licensing costs became too high relative to expected economic returns. This reflects fears that overly restrictive copyright rules could discourage investment and reduce the UK’s competitiveness in the global AI market – the opposite intended outcome of the UK AI strategy.
- Option 3: text and data mining exception. The proposal for a broad TDM exception with no opt-out mechanism was seen as creating significant risks for the creative industries. The assessment concluded that such an approach could reduce incentives to invest in creative output by weakening creators’ ability to control and monetise their work. While this option could increase the competitiveness of the AI industry, it risked doing so at the expense of creators and rightsholders.
- Option 4: text and data mining exception with opt-out clause. This option attempted to strike a compromise between innovation and creator protection, and seemed to gain the most traction in the early stages of the consultation. However, the consultation results demonstrated very limited support for this approach, with it receiving only a 3% approval rating. The UK Government has therefore rejected it.
Summary of the outcomes
In short, the UK Government has decided not to do anything yet.
Every option involves trade-offs between supporting AI innovation and protecting the economic interests of rightsholders: the UK Government is unwilling to commit to favouring one interest over the other.
The Government’s 2026 potential UK AI strategy
The Government’s indecisive response suggests a more cautious and incremental UK AI strategy moving forward. Rather than committing immediately to major legislative reform, the government identified four key areas of focus for the coming months.
Digital replicas consultation
First, it plans to launch a consultation on digital replicas in summer 2026. This is likely to address growing concerns around AI-generated imitations of performers, creators and public figures, including voice cloning and synthetic likenesses.
We previously covered changes to Danish law to bring in more protections for individuals who have been “deepfaked.” It is possible that similar legislation could be introduced in the UK.
Taskforce for labelling AI-generated content
Second, the government will establish a taskforce for labelling AI-generated content, with an interim report expected in autumn 2026.
This reflects broader concerns around transparency, authenticity and consumer trust in an environment where AI-generated media is becoming increasingly difficult to distinguish from human-created content.
Review control of creators over their publicly available works
Third, the government will publish a review examining the mechanisms available for creators to control their works online. This review will consider standards, technical solutions and best practice regarding input transparency for AI model training.
Review government involvement in licensing content
Fourth, the government will launch a working group focused on independent and smaller creative organisations. The group will explore whether government has a role in supporting these organisations’ ability to license content effectively in an AI-driven market.
Alongside these initiatives, the government also announced the establishment of the Creative Content Exchange (CCE). The CCE is intended to function as a secure and trusted online marketplace for licensing digitised cultural and creative assets, with an operational pilot platform expected by summer 2026.
The initiative suggests that the government sees licensing infrastructure as a potential key aspect of the UK AI strategy to at least some of the tensions between AI developers and rightsholders.
A Deliberately Cautious Approach to AI and Copyright
According to the UK government, the current evidence base remains too limited and the opinions on this matter remain too varied to take any decisive course of action in relation to the impact of copyright law on the development and deployment of AI within the UK.
This is a notable admission. Despite the intensity of the public debate, the government appears to recognise that many of the long-term economic consequences of AI-copyright reform remain unclear.
As a result, the current direction of travel is one of continued evidence gathering, consultation and technical exploration rather than immediate legislative change.
How AI Developers and Rightsholders Should Act Now
Being uncertain of impending legislative change is a frustrating position for AI developers and rightsholders alike. The debate surrounding AI and copyright is set to continue for the coming months, potentially even years, and the recent consultation only confirms that the UK government is not nearing any decisive action.
Relevant individuals and organisations should therefore keep a keen eye on the numerous consultations, task forces, and reviews that the government will be actioning in the coming months.
AI developers should similarly review how they are training their models, and prepare to justify their use of copyrighted works by purchasing licenses, posting disclaimers, and putting up safeguards if necessary.
Under current laws, the most certain position is to obtain a licence – that is unlikely to change in the short term. AI developers can act now to obtain licences to obtain much more confidence in the material that they are using. Likewise, rightsholders can engage directly with AI developers, publishers or collective licencing organisations in order to support the licencing of their creative works.
At EM Law, we are experts in the legal issues surrounding AI and intellectual property. If you need help navigating AI or intellectual property, please contact us here or visit our AI Lawyers or Software & Tech Lawyers pages for more information.
