UK maps the bill if America cuts off frontier AI
Two months after Washington yanked foreign access to Anthropic's newest models, London is finally asking what that would cost.
Britain's Cabinet Office has asked the ministry that oversees business, science and trade to measure the economic hit if the UK is locked out of the latest frontier AI models, the Financial Times reports, citing people familiar with the matter. The review is expected in a matter of weeks. Sina Finance, carrying the same account, says the request was prompted by a June White House order that pulled foreign users off Anthropic's Fable 5 — a shock that left the National Security Secretariat worrying the next cutoff would not be temporary. The government has not published a brief, and the people talking to the FT are unnamed, so treat the commission as reported rather than confirmed from Whitehall. The fact that the question is being asked at all is the news.
That June episode was not a thought experiment. The Telegraph reported at the time that the Trump administration ordered Anthropic to restrict Fable 5 and Mythos 5 to Americans only, a move UK officials called a disaster. Access came back in pieces. The lesson did not: frontier models are not a subscription a treasury can budget for. They are a permission slip issued in Washington, and it can be torn up in an afternoon. Once you have seen the switch thrown, you stop talking about "the cloud" as if it were weather.
The review is the government admitting as much in public. Hosting servers is not the same as controlling them. Oxford researchers told Parliament in July that the UK sits on about 4 percent of the world's AI compute and exercises sovereign control over only about 0.1 percent of it — meaning machines on British soil can still go dark if a US cloud provider, or the US government, says so. Associate professor Amro Awad put it bluntly: if those companies or their governments restrict access, "the physical servers in UK data centres could become unusable." Professor Robert Trager, who directs the Oxford Martin AI Governance Initiative, framed the political question the same way: whether Britain has "reliable, sovereign access to the frontier of artificial intelligence." An assessment of "what if we lose access" is the policy version of checking whether the spare key still works.
The take is that London is pricing a risk it already lived. Britain's sovereign-AI programme was sold as insurance against exactly this kind of dependency. An economic review does not train a model, and it does not negotiate a guarantee from Washington. What it can do is put a number on dependence, which is the one argument that still moves a Cabinet. If the figure is large, expect a louder push for access treaties, inference sovereignty, and open-weight hedges. If it is small, expect Whitehall to keep renting the frontier and hoping the next Fable 5 moment is someone else's problem. Either way, the review is an admission that "special relationship" is not an SLA.
What to watch: whether the review publishes a number — and whether London asks Washington for a written access guarantee before the next model ships.
If America can switch the UK off in an afternoon, is a cost study the start of a plan — or a way to postpone one?
