Alan Turing Institute chief warns UK against relying on foreign AI

The UK must build its own artificial intelligence capability rather than relying on foreign technology that could be restricted or turned off without warning, the head of the Alan Turing Institute has warned.

George Williamson, chief executive of Britain’s flagship AI research body, said that depending entirely on overseas systems creates severe risks for national resilience, particularly as AI becomes embedded in healthcare, defence, national security and energy networks.

His warning follows recent friction in international technology supply chains, including Washington temporarily blocking the export of Anthropic’s latest AI models outside the United States in June. US authorities have also reportedly asked American AI labs to withhold new models from the UK’s AI Security Institute until US regulators examine them first.

Williamson said political and commercial pressures mean Britain cannot guarantee uninterrupted access to foreign models during a crisis. Relying on software that the UK cannot inspect, control or run without permission puts essential services at risk, he argued.

Why sovereign AI matters for Britain

While Williamson acknowledged that the UK cannot match the financial firepower of American tech giants or the Chinese state, he argued that Britain should develop its own intellectual property and smaller, specialized models. These could be made available to domestic startups and public services.

He highlighted practical uses for national models in agriculture, weather forecasting and tracking state-backed disinformation campaigns. Williamson also noted concerns over Chinese AI systems, pointing out that their outputs reflect political censorship that does not align with British democratic values.

Williamson took over the Turing Institute after ministers pushed for a strategic shift in its work. In 2025, Peter Kyle, then technology secretary, instructed the institute to prioritize defence and security, leading to the resignation of Williamson’s predecessor, Jean Innes.

A former head of His Majesty’s Government Communications Centre, Williamson said framing the institute's work around national resilience allows it to address both defense threats and environmental challenges, calling climate change a national security issue on par with defense.

Assessing AI risks and existential claims

Williamson also expressed skepticism about public declarations regarding existential risks from AI, such as claims that advanced systems carry a significant chance of causing human extinction by 2030.

He described percentage-based forecasts of existential threat as poorly grounded in science, stating that he has seen no convincing evidence explaining how those numbers are calculated. A national institute, he said, should serve as an objective guide to help government and the public navigate genuine operational risks without getting distracted by speculative figures.