Alan Turing Institute chief urges UK to build AI it can control

Britain needs AI it can use without another country’s permission—but that need not mean matching the biggest labs’ spending. In a Guardian interview published October 10, Alan Turing Institute head George Williamson called for stronger domestic capability and outlined a national-resilience agenda spanning critical services, AI threats and public-good research.

Control without copying the biggest labs

Williamson’s concern is not simply where an AI company is headquartered. He warned against relying solely on systems Britain cannot inspect or control, particularly as AI becomes embedded in defence, national security, critical infrastructure and healthcare. Access could be restricted or switched off, he said.

Political and commercial pressures have shown that access may not always be available when the UK wants it, Williamson argued. He also questioned how the country could build an economy around technology that might disappear. His proposed answer is greater domestic control over access, rather than dependence on permission from others.

That concern has a recent backdrop. The Guardian reported that Washington temporarily blocked releases of Anthropic’s latest models outside the US in June. It also reported that US officials had asked leading labs to withhold models from Britain’s AI Security Institute until the US counterpart had examined them.

Williamson acknowledged that the institute and the UK probably could not match major AI labs’ financial resources. Instead, he suggested developing nationally owned intellectual property or smaller models that startups could use. He presented these as ways to build domestic capability without replicating the work of the leading labs.

He also raised concerns about whether Chinese-made systems reflect UK values. Pointing to answers about recent Chinese history, he argued that Britain should help shape AI in ways that accord with its own values—not merely secure access to someone else’s technology.

We can be ambitious here without trying to copy what frontier labs are doing.

George Williamson, speaking to the Guardian

Resilience brings climate back into the frame

At the institute, Williamson said national resilience would mean securing critical systems against disruption, countering threats from AI itself and helping shape the UK’s strategic AI capability. He framed those priorities as a response to a more adversarial world, repeated infrastructure shocks and powerful AI that can amplify threats.

His definition also includes public-good applications, from weather forecasting to farming. Monitoring state-led disinformation is another area where he said the institute could help, describing those campaigns as dangerous for democracy. The agenda therefore reaches beyond protecting systems to using AI against risks facing the country.

That breadth follows pressure for a narrower remit. In 2025, then technology secretary Peter Kyle wrote to the institute’s chair demanding that its main focus shift to defence and security. The government is its main funding source, and Williamson’s predecessor, Jean Innes, resigned soon after that demand.

Williamson said the resilience framing would allow environmental work to continue. “For me climate is as much a national security issue as defence,” he said. He described resilience as a frame broad enough to cover the range of threats and opportunities facing the UK.

A scientific role in the risk debate

Williamson also sees a role for the institute, a registered charity, as an objective arbiter of AI concerns, including existential threats. He said a national institute should help both government and the public navigate those questions, while warning that some discussion of extinction risk is not well grounded in science.

His specific objection was to assigning percentages to the chance of AI-triggered extinction. Williamson said he had not seen convincing evidence explaining how people arrived at those figures. He questioned the usefulness of numerical estimates, rather than arguing that the institute should ignore existential risk.