A wave of stark warnings from computer scientists, an AI industry insider and politicians this week has intensified calls for tighter controls on the development of artificial superintelligence (ASI), technology that could surpass human capabilities. In Westminster, figures including former defence secretary Des Browne and Berkeley professor Stuart Russell compared the risks to nuclear weapons, while an Anthropic researcher's resignation and a colleague's admission of a "greater than 10%" chance AI could "kill all humans" within a decade have added urgency to demands for government intervention on both sides of the Atlantic.

The Westminster session, convened by lobbying group Control AI, backs a bill from Labour MP Alex Sobel seeking to ban ASI creation, while Anthropic's Evan Hubinger acknowledged his firm lacks a plan to ensure such systems remain safe, after researcher Jacob Coxon quit accusing Anthropic and OpenAI of "gambling with our lives". Concerns were also raised that Anthropic had not submitted its newest model, Mythos 5.1, for UK pre-release safety testing. Labour MP Darren Jones has written to the prime minister and international bodies urging a multinational safety treaty, while in the US, senator Bernie Sanders cited polling showing 81% of Americans want Congress to regulate AI.