New UK AI plans: Labour throws its hat into the AI ring

Yesterday was a busy day for AI in the UK. Kier Starmer launched the Government’s plan to make the UK a ‘great AI superpower’, publishing both Matt Clifford’s AI Opportunities Action Plan and the Government’s response to it.  Within days of taking office, he had commissioned venture capitalist Clifford to develop a plan ‘for harnessing AI’s boundless potential’, and the result is an ambitious, wide-ranging plan containing 50 recommendations for the UK Government to:

• grow the UK’s AI sector;
• drive adoption of AI across the economy to boost growth; and
• improve products and services.

Government response:

In its response, the Government emphasises its commitment to build on the UK’s strengths in this area and maintain its position as a global leader, despite increasing competition. It also stresses the importance of acting now, given the fast pace of technological change, as well as AI’s potential to “turbocharge every mission in this Government’s Plan for Change”. 

The Government agrees (either wholly or partially) with all of the plan’s recommendations (with a few caveats*, and – of-course - acknowledging that the devil is in the detail). 

The recommendations cover a range of areas from infrastructure and investment to access to data, attracting talent, AI regulation/assurance and AI use in government. Examples include:

• Building sufficient, secure and sustainable AI infrastructure: this includes a commitment to expand the UK’s sovereign compute capacity by at least 20x by 2030, and build a new state of the art supercomputing facility (something the previous Conservative Government had proposed, but which Labour postponed due to funding concerns). The Government will also publish a long term “Compute Strategy” in Spring 2025 and set up ‘AI Growth Zones’.
   
• Unlocking data assets in the public and private sector: AI developers need access to high quality data. The creation of a National Data Library (together with the Government’s wider data access policy) will allow the Government to ‘responsibly, securely and ethically’ unlock the value of public sector data assets to support AI research and innovation. Although this has already raised a number of privacy concerns, the Government’s response states that the plans will all be underpinned by ‘strong privacy-enhancing technologies’. 
   
• Enabling safe and trusted AI development and adoption through regulation, safety and assurance: The government sees the UK’s current pro-innovation approach to regulation as “a source of strength relevant to other more regulated jurisdictions and the UK should be careful to preserve this” (presumably code for the UK will not introduce an EU style AI Act). But what does this mean in practice:
  • The Government will “set out its approach to AI regulation.” DSIT will consult on proposed legislation “to help kickstart growth and protect UK citizens and assets from critical risks associated with the next generation of the most powerful AI models” – something originally mentioned in the King’s speech (see blog). It will also establish the AI Safety Institute as a statutory body.
  • The Government will “act to ensure that [the UK has] a competitive copyright regime that supports both [its] AI sector and the creative industries”. This has proved a difficult balancing act in the past, with failed attempts by both successive Governments and the UK IP Office to resolve the issue. *While the Action Plan recommends reforming the UK’s text and data mining regime so that it is at least as competitive as the EU, the response merely references the fact that the UK is already consulting on this (see blog).
  • Regulators, the Government says, have an important role to play in supporting innovation “as part of their growth duty”.  While the Action Plan recommends a commitment to fund regulators to scale up their AI capabilities, and the Government response agrees with this recommendation, the detail around this is less certain. Sponsor departments will liaise with regulators to identify needs, taking into account existing funding models, which DSIT will then assess with HM Treasury support. There are a number of other recommendations relating to regulators, including that those with significant AI activities will publicly report on their actions to promote AI innovation on an ongoing basis, that the Regulatory Innovation Office (‘RIO’) may provide targeted strategic guidance to regulators and that DSIT, through the RIO, will work with regulators to identify pro-innovation initiatives.    

Next steps and comment:

The Government seems keen for these plans to turn quickly into actions, with its response including a detailed table listing each Action Plan recommendation, with the Government response to it and a delivery timeline - and a unit based in DSIT has been tasked with reporting on progress. The Government has said it will continue to develop its response to the Action Plan ahead of the Spring 2025 spending review and will include its wider approach to AI in the upcoming Industrial Strategy’s Digital and Technologies sector plan. 

Commenting on the plans, co-head of our Technology Group Rob Sumroy said: “The UK has for a number of years now taken an innovation-friendly approach to AI. Despite the change in Government, these plans build on that approach. While there had been much focus in the last year or so on the safety aspects around AI, I don’t see the new proposals as ignoring those concerns - rather they champion the importance of the UK securing investment around AI, building the necessary infrastructure and skills and leveraging its pragmatic regulators to foster a culture of AI innovation.