Vallance to Chair UK’s New Government-Wide AI Taskforce

The UK has moved artificial intelligence to the centre of its government, standing up a Prime Minister’s AI Taskforce that reports straight to Downing Street and pulling the country’s frontier-model testing body into the Cabinet Office alongside it.

Prime Minister Andy Burnham announced on July 24, 2026 that Lord Vallance will chair the new unit. Vallance, the government’s chief scientific adviser from 2018 to 2023 and its science minister until this week, will report directly to the prime minister.

A delivery unit, not a regulator

The Taskforce’s remit is adoption, not rulemaking. The government says it will set and implement the state’s overall AI strategy and push the technology into public services, with the stated aim of unlocking growth “across the country.” Ministers modelled it on the Vaccines Taskforce that Vallance ran during the pandemic, an empowered team with a single line of accountability to the prime minister.

The structure carries more weight than the branding. The Taskforce will sit inside a new Office for the Prime Minister and the Cabinet, next to No.10 and core Cabinet Office functions, reporting to Cabinet Secretary Antonia Romeo. The government is billing the shake-up as the first time AI has been placed “at the very centre of government,” with Kanishka Narayan, the minister for AI, running the unit day to day and now taking a seat at the Cabinet table. Vallance also picks up the chairmanship of the Oxford-Cambridge Growth Corridor, the government’s flagship innovation-cluster project.

What the announcement does not add is any new statutory power. The UK still has no binding AI law to match the EU’s AI Act, the risk-based regime now phasing obligations onto general-purpose and high-risk systems. Successive UK governments have preferred a lighter, sector-by-sector approach, and this restructuring reinforces that path: AI is being treated as an instrument of industrial strategy and public-service reform rather than as something to be regulated from the centre.

The safety institute moves with it

The most consequential line for the wider market is procedural. Responsibility for the AI Security Institute, the state-backed body that runs safety and security evaluations on frontier models, will transfer to the same Office for the Prime Minister and the Cabinet.

The institute, which calls itself the first state-backed organisation of its kind, has built a technical team of more than 100 staff and works with leading AI developers, including Google DeepMind, on evaluating frontier systems. Its assessments feed international coordination on model risk and, increasingly, decisions about which systems governments will buy. Shifting its reporting line out of the science department and into the prime minister’s office reframes model evaluation as a centre-of-government function rather than a specialist scientific one.

It also raises a question the announcement leaves open: how a body whose authority rests on technical independence will be positioned inside a unit whose explicit job is to accelerate AI adoption and growth.

A reshuffle that rewired Whitehall

The changes land days into a new administration. Burnham replaced Keir Starmer as prime minister on July 20, 2026, and the reshuffle that followed folded the standalone science department into a broader economic ministry and ended Vallance’s spell as science minister on July 22, 2026. Handing him a PM-level AI brief brings him back with a narrower, delivery-focused mandate.

For companies that build and sell AI, the practical shift is about where decisions now sit. Public-sector procurement, the adoption agenda, and the evaluation regime that shapes which frontier models are judged acceptable are being consolidated under one roof answering to the prime minister. It is a bet that the same centralised, deadline-driven approach that delivered vaccines can move a technology whose governance questions, from copyright to model safety, remain far from settled.

That stance sets London further apart from Brussels, where a binding rulebook is already phasing into force, and reflects a wider contest among middle powers trying to turn AI capability into national advantage. Whether concentrating the agenda in Downing Street speeds up public-service transformation or simply crowds it will depend on what the Taskforce ships.