On 13 January 2025, the UK's Prime Minister, Sir Keir Starmer, made a speech designed to showcase the new AI Opportunities Action Plan. This was commissioned by the Government and led by Matt Clifford CBE, a tech entrepreneur and the Chair of the Advanced Research and Invention Agency (ARIA). Containing 50 recommendations for the Government to implement, the Action Plan's broad objective is to make the UK "one of the great AI superpowers", building on its existing status as the third largest AI market in the world, behind the USA and China.
The Action Plan, the Government's response to which is accessible here, gives rise to significant implications for AI developers, rights holders, and individuals as regards intellectual property, data protection, and the overall direction of travel in the UK concerning the regulation of AI-powered technologies.
What does the AI Opportunities Action Plan propose?
Key takeaways from the Action Plan include the following:
- Expanding the capacity of the AI Research Resource (AIRR). The Action Plan envisages increasing the capacity of the AIRR—a cluster of advanced supercomputers for AI research—twentyfold by the end of the decade. It recommends the establishment of AI Growth Zones (AIGZs), areas with enhanced power access and fast-tracked planning regulations intended to accelerate the spread of AI infrastructure in the UK. The first is intended to be in Culham, Oxfordshire. Already home to the UK Atomic Energy Authority, the site is earmarked for the development of a pilot AI data centre beginning with 100MW of capacity, potentially scaling up to 500MW.
- Attracting AI talent. To bridge the estimated gap between supply and demand of skilled AI professionals, new educational pathways will be explored. While the Action Plan also advises of changes to the existing immigration system to attract the required talent, the Government's response does not indicate full agreement with this aspect of the recommendation, stressing that the UK already offers a variety of visa routes for this purpose.
- Adopting AI in the public sector. Large-scale adoption of AI within the public sector is recommended not only to maximise productivity but also to encourage private-sector adoption. Examples include medics using AI to complete forms and reports and teachers using these technologies to plan lessons. A range of use cases are to be piloted, with the best deployed nationwide.
- Unlocking the UK's data assets. This includes revisiting the UK's legal framework concerning copyrighted-protected assets that might be used by AI companies to train AI models and creating a new National Data Library (NDL) that comprises public-sector data, potentially including health-related data from the NHS. The Action Plan additionally recommends establishing a copyright-cleared UK media asset training data set for licensing internationally, and which might include content from institutions such as the BBC, the National Archives, and the British Library.
- Adopting a pro-innovation approach. Although the Action Plan initially refers to this particular principle in terms of "enabling safe and trusted AI development and adopting through regulation, safety and assurance", it is at pains to emphasise opportunities for maximising economic growth and innovation, urging the Government to ask itself whether each recommended action "benefit[s] people and organisations trying to do new and ambitious things in the UK" and warning that, if the answer to this is negative, then untapped potential may not be realised.
How will this affect AI regulation in the UK?
While the Action Plan reaffirms support of the UK's Artificial Intelligence Safety Institute, it does represent a shift in emphasis towards underlining AI technologies' potential to boost economic growth. The Government's broad agreement to the Action Plan indicates that the UK will not pursue wholesale regulation of these technologies on a statutory basis. With the UK no longer part of the EU, it understandably makes little political sense for the Government to adopt an approach equivalent to the EU AI Act, the world's first comprehensive and legally binding framework for AI development and use. Additionally, Britain's sluggish economic growth means there is little appetite to create further deterrents to potential investors by way of regulation. It may be no coincidence that the Prime Minister announced the Action Plan on the same day the pound hit a 14-month low against the US dollar.
The Action Plan reinforces its pro-innovation agenda by going as far as to state that regulators entrusted with oversight should be asked to report on how they have helped to foster AI-driven growth. Further, it adds that if those reports demonstrate that regulation is in fact stifling innovation, the Government should consider transferring the relevant regulatory powers to a central body with "higher risk tolerance".
It is therefore clear that, while alignment with the EU's approach regarding text and data mining (as discussed below) may be explored by the Government, any attempt to mimic the EU AI Act in the UK is now unlikely. Nonetheless, UK businesses wishing to export AI technologies to the EU will still be bound by the requirements of the EU AI Act, whose effects may, in this way, still be felt in Britain.
What does this mean for the UK's intellectual property (IP) framework concerning content used to train AI models?
There is an ongoing tension between the need for AI developers to use a significant number, and broad range, of materials as training data for their models and the interests of IP rights holders and creators. The current text and data mining (TDM) exception in UK copyright law only covers the use of copyright-protected materials for non-commercial research, and the previous Conservative Government's attempts to broaden its scope (ie a broad TDM exception with no opt-out for rights holders) were abandoned, with attempts to broker a voluntary resolution subsequently failing.  
As discussed in our December article, the UK Government has issued a fresh consultation on this subject, proposing a TDM exception for commercial purposes, subject to an opt-out right by rights holders. Although stakeholders can contribute to the consultation until 25 February 2025, the Action Plan considers the current position unsatisfactory, claiming it is "hindering innovation and undermining our broader ambitions for AI". This indicates a clearer desire for the UK to be at least as competitive as the EU as regards making data sets accessible to AI developers.
The Government's published response to this aspect of the Action Plan simply states that the issue is open for consultation. However, the fact its own publicity materials generally refer to the Prime Minister "agreeing to take forward all 50 recommendations" has already left some rights holders concerned as to whether responses to the consultation will be heeded. Within hours of the Action Plan's publication, for example, the CEO of the Publishers Association cautioned that the outcome of the consultation should not be a fait accompli, adding his opinion that "[t]he UK can […] seize all the growth opportunities associated with AI without facilitating a US tech-led heist of UK copyright work".
Does this address the use of NHS health data in an AI context?
The NHS is one of the world's largest single sources of patient-level health data. It is no surprise, therefore, that its data assets are very attractive for AI developers that have already exhausted various pools of training data.
While this is one opportunity that affords the UK a unique selling point compared with other competing economies, any move which might involve large technology companies accessing NHS data is not without political or legal risk.
To alleviate these concerns, the Action Plan envisages holding the relevant data sets in the NDL, with the Prime Minister clarifying in his speech that such data would be anonymised. If the data is truly anonymised such that no individual can be re-identified, its use would fall outside of the scope of data protection law. Nonetheless, in response to the Government's announcement, the Information Commissioner's Office has stressed that "[d]ata protection is essential to realising this opportunity and ensuring that the public can have trust in AI".
There also remain questions regarding the feasibility of some of the proposals, not least in terms of how some aspects—such as the expansion of the AIRR's capacity—will be funded given the current economic climate. However, as mentioned above, the Government's response to the Action Plan indicates that it broadly agrees with all recommendations.