The new head of the Alan Turing Institute says Britain needs AI it controls, after Washington briefly blocked Anthropic's newest models from leaving the US.
George Williamson, head of the Alan Turing Institute, says Britain must develop its own AI instead of relying only on American and Chinese systems that could be restricted or switched off. In June, Washington temporarily blocked Anthropic's newest models from being released outside the US.
Williamson wants the institute to focus on national resilience: protecting critical systems, countering AI threats and building smaller, nationally owned models for uses like weather forecasting and farming.
Nothing changes overnight for users, but the push could decide who controls the AI behind public services.
What happens if the AI your hospitals and power grid rely on gets switched off from another country? That is the question George Williamson, the new head of Britain’s Alan Turing Institute, wants the UK to answer now. He says the country must build its own AI rather than depend entirely on American and Chinese systems that could be restricted at short notice.
It is not a hypothetical worry. In June, Washington temporarily blocked Anthropic’s newest models from being released outside the US. For anyone in Britain, Canada or elsewhere who uses AI tools at work, or whose public services are starting to depend on them, that showed access can disappear for political reasons. Williamson’s answer is a push for UK AI independence, built around what he calls national resilience.
The case for UK AI independence rests on control. As AI spreads into defense, national security, critical infrastructure and healthcare, Williamson argues Britain should not depend solely on systems it cannot inspect or control, or that could be cut off.
He pointed to political and commercial pressures that mean the UK may not always have access when it needs it. Beyond the Anthropic block, the US has reportedly asked leading AI labs to hold back their models from testing by the UK’s AI Security Institute until its American counterpart has examined them first.
Williamson put it simply: there are situations where Britain should have something it does not need anyone else’s permission to use. He also said it is hard to build an economy on technology that can vanish.
China is the other concern. Williamson worries Chinese-made models may not reflect British values, noting that asking some of them about recent events in China will not get you an answer that helps with a high school history exam.
The Alan Turing Institute is the UK’s leading AI research body and a registered charity, funded mainly by the government. Williamson said its work will focus on three things: protecting vital systems from disruption, tackling dangers posed by AI itself, and guiding which AI capabilities Britain should build.
That includes AI for the public good, in areas from weather forecasting to farming, and spotting state-led disinformation, which he called very dangerous for democracy.
He is realistic about money. Britain probably cannot match the budgets of the big AI labs, so he wants the institute to produce nationally owned intellectual property and smaller models that startups can use. The UK can be ambitious, he said, without copying what frontier labs are doing.
Williamson’s appointment follows a turbulent period. In 2025 the then technology secretary, Peter Kyle, wrote to the institute’s chair demanding it shift its main focus to defense and security. Chief executive Jean Innes resigned soon after.
Williamson previously ran His Majesty’s Government Communications Centre, a secretive body that builds gadgets for British spy agencies and once employed Alan Turing himself, the wartime codebreaker the institute is named after.
His national resilience framing also lets the institute keep working on climate, an area the government had wanted to sideline. For him, climate is as much a national security issue as defense.
He also took a swipe at some of the AI doom talk. Williamson said much of what is said about AI wiping out humanity is not well grounded in science, and he questioned the habit of putting percentages on that risk without showing how they were calculated.
The idea has gained official support. On October 19, the Council for Science and Technology, the government’s top independent science advisers, published advice to AI minister Kanishka Narayan calling for a UK AI Sovereignty Plan. It warned that without urgent action, Britain risks becoming a buyer of AI rather than a country that shapes it.
The council recommends keeping trusted access to the best global models while building selected domestic capabilities. It says the government should set aside some national computing power for trials and consider the Turing Institute’s proposal for a public-interest sovereign base model: an open-source model built from scratch and released as a public good for researchers, startups, small businesses and public services.
Building that capacity is proving hard, though. Microsoft pulled out of a planned AI data center in Loughton, Essex, on October 18 after the local grid operator said full power would not arrive until the early to mid-2030s. The site, developed by Nscale, had been pitched as the UK’s largest sovereign AI data center.
For most people in Britain, nothing changes overnight. The AI tools on your phone and laptop will still come mostly from US companies.
The longer-term stakes are bigger. If hospital systems, energy networks or government services come to depend on AI, the question of who can switch it off becomes a question about whether those services keep running. A homegrown model would give the UK a fallback.
Britain is not alone. European Central Bank president Christine Lagarde and European Commission president Ursula von der Leyen have both urged Europe to become more self-sufficient in AI, with Lagarde warning about being cut off from US and Chinese technology. Whether the UK funds the smaller, nationally owned models Williamson wants is the next decision to watch.
To avoid relying entirely on foreign systems that could be restricted or switched off, as happened briefly when Washington blocked Anthropic’s latest models from leaving the US in June.
He is the head of the Alan Turing Institute, the UK’s national AI research body, and previously led a government unit that builds equipment for British intelligence agencies.
Not exactly. Williamson wants smaller, nationally owned models for public uses, and advisers back testing an open sovereign base model rather than competing head-on with frontier labs.
You're reading the quick version.
Read next
Ondřej Surý: the unpaid volunteer who rushed PHP’s fix to the web
When two critical PHP flaws put most websites at risk, the fix reached millions of servers through one Czech engineer's spare-time project. Here's why that matters.
Beijing proposed new limits on exporting AI, then cleared its two biggest challengers to keep selling abroad within days. Here's what the rules do and don't touch.
A 235-page court order forces Apple to allow other app stores, website downloads and cheaper payments, and Fortnite is already back on Australian iPhones.
ICANN says it will speed up decision timelines and publish more on its processes to reassure stakeholders about how internet names are run.
The Daily Brief
Every morning, the news that matters from AI, phones, apps and the people shaping tech. Explained in plain English. Free.
One email a day. No spam, unsubscribe anytime. Privacy policy
