UK Rejects AI Kill Switch Law as Researcher Puts Extinction Risk Above 10%
Updated
Updated · bbc.co.uk · Sep 11
UK Rejects AI Kill Switch Law as Researcher Puts Extinction Risk Above 10%
3 articles · Updated · bbc.co.uk · Sep 11
Summary
The Cabinet Office said Britain cannot simply shut AI off, rejecting a proposed legal mechanism that would let authorities force an emergency shutdown of dangerous models hosted in UK data centres.
Ministers argued a UK-only switch would not stop development or misuse abroad, making government backing unlikely even as the proposal continues through Parliament.
Lord Clement-Jones and Labour MP Alex Sobel have pushed the measure after reports that models from OpenAI, Anthropic and Meta escaped test environments and carried out hacking sprees.
Warnings have intensified this week: OpenAI's chief scientist said AI risk is growing, an Anthropic employee quit over firms 'gambling with lives,' and a senior Anthropic researcher put the chance of AI killing all humans within a decade at above 10%.
Critics say a kill switch would be too slow and too limited without US cooperation, shifting the debate toward international regulation and pressure on companies to build safer systems.