The Council for Science and Technology on October 19, 2026, published a letter of advice to Minister for AI Kanishka Narayan on how the UK can build and direct its own AI capability, setting out four recommendations aimed at what the Council calls AI sovereignty.
The letter is dated July 30, 2026, and signed by the Council’s co-chairs, Dame Angela McLean and Lord Browne of Madingley, who congratulated Narayan on his appointment. The co-chairs wrote that the Council is developing advice for the government on AI sovereignty and attached a summary of that advice as it currently stands. The Council describes itself as the government’s top-level independent advisory body on cross-cutting science and technology issues, drawing together leaders from academia, business and finance to advise the Prime Minister and the Cabinet.
The letter defines AI sovereignty as having “sufficient agency across the AI value chain to deliver the UK’s chosen outcomes, capture economic and public value for the UK, and maintain resilience under duress.” The Council wrote that choices made now will shape the UK’s long-term competitiveness, resilience, public trust and strategic autonomy, and that the benefits of early action are substantial. It warned that without deliberate and urgent action, the UK risks becoming a buyer of AI rather than a country able to shape the technology and how it is used.
The letter was copied to the Prime Minister, the Secretary of State for Business, Innovation, Science and Trade, the Secretary of State for Defence, the Secretary of State for Digital, Culture, Media and Sport, the Chancellor of the Exchequer, the Chancellor of the Duchy of Lancaster, the chair of the Prime Minister’s AI Taskforce, the Cabinet Secretary and relevant departmental permanent secretaries. The co-chairs credited Council members Professor Sir Nigel Shadbolt, Avid Larizadeh Duggan, Mark Enzer, Priya Lakhani, Sir John Lazar, Professor Nick McKeown, Professor Andrew Morris, Professor Sarah Sharples and Richard Slater with developing the advice.
Ecosystem Mapping, Pilots and Governance
The first recommendation calls for a UK AI Sovereignty Plan built on a rapid mapping of the main parts of the AI ecosystem: computing power, data, skills, models, governance and public adoption. The map should identify existing activity, dependencies and gaps, and the Council said it should be produced quickly and used as a practical decision tool to determine which capabilities the UK should own, which it should pursue in collaboration with others, and which it should secure resilient access to.
Those choices should inform a set of rapid AI pilots applied to public-interest use cases such as national security, health or education, each with engagement plans that give the public a role in shaping the design. The Council said every pilot should answer six questions: what problems are to be solved, what sovereign data the UK holds, what model capability is needed, what system must be built around it, what value can be captured, and how success will be measured.
The second recommendation covers governance and coordination. CST advised the government to designate a single accountable owner, such as the head of the Cabinet Office AI Taskforce, to bring departments together and shape funding and delivery choices. Governance, the letter says, should combine strategic and technical advice to direct the pilots and assess model, data and computing options.
Skills, Data and a Sovereign Base Model
The third recommendation urges immediate investment in the foundations of long-term UK AI capability, which the letter describes as not only models but people, skills, organisations, data, trust and adoption. The Council called for attractive career pathways and opportunities for practical experience so that expertise stays and grows in the UK, an AI skills and secondment route that complements TechFirst, a stronger evidence base on AI adoption and productivity, and continued support for fundamental computing research including low-power approaches.
The letter describes data as one of the UK’s most important assets, valuable only when managed, secured and used effectively. CST recommended that the government identify important UK datasets and test access arrangements in at least two high-value areas, such as health and creative data.
The fourth recommendation says the UK should maintain trusted access to leading global AI models while building selected domestic capabilities that strengthen independence, resilience and future choice. Access to the most powerful global models remains important, the Council wrote, but the UK should not depend on them for every use case, and in some cases a more specialised system developed domestically will work better.
The pilots, CST said, should test how different types of AI systems, such as open-weight models and smaller domestic models, address different requirements and tasks, and trialling different ways to build, train and access models will build enduring national skills, institutional knowledge and technical judgement. The letter says the government should allocate some national AI computing capacity for these trials and should consider proposals such as the Alan Turing Institute’s for a public-interest sovereign base model. A footnote describes that proposal as an open-source sovereign base model built from scratch and released as a public good: an open, UK-governed asset available to researchers, start-ups, small and medium-sized enterprises, public services and the open-source community, rather than a proprietary product for a single firm or department.
The Council said it can provide further independent support and advice as required.
