UK and US Leaders Clash Over AI Kill Switches

The UK Cabinet Office and U.S. President Donald Trump have rejected proposals to mandate emergency kill switches for powerful artificial intelligence systems. The UK government dismissed an amendment to the Cyber Security and Resilience Bill proposed by Lord Tim Clement-Jones, arguing that domestic shutdown mechanisms are infeasible because models could simply be hosted elsewhere.

In the United States, lawmakers introduced the Kill Switch Act to require shutdown controls for problematic tools. However, Donald Trump dismissed warnings about AI existential risks as a "hoax" and a "sick conspiracy" designed to allow China to maintain a technological lead. He argued that the U.S. must prioritize its competitive edge in the AI race.

Industry figures, including Anthropic co-founder Jack Clark, continue to advocate for verifiable shutdown mechanisms. Clark stated that society may eventually require rules ensuring these switches are inspectable by independent third parties. This push follows warnings from AI pioneer Geoffrey Hinton, who described a 10 percent chance of AI eliminating humanity as "not unreasonable."

While some researchers and leaders from OpenAI and Anthropic have agreed to external reviewers, others remain skeptical. Former OpenAI researcher Daniel Kokotajlo supported the UK government's rejection of domestic switches, arguing that users would still be "steamrolled" by super intelligences developed in the United States.