Lords Call It: AI Needs a Kill Switch Now

The UK House of Lords has passed a legislative amendment requiring the government to develop emergency "kill switch" powers over advanced artificial intelligence systems, according to the BBC — a move that would give British authorities the legal mechanism to shut down or suspend AI operations deemed to pose an unacceptable risk to public safety.

The amendment, backed by a coalition of peers citing catastrophic-risk scenarios, specifically names systems of the scale developed by OpenAI and Anthropic as the category of concern. Supporters argue the provision would act as a "vital safety net" against AI systems that escalate beyond human oversight — not a ceiling on innovation, but a circuit breaker. Critics warn the threshold for activation remains undefined, leaving the mechanism open to politically motivated use.

The vote lands as AI regulation in Europe and the United Kingdom diverges sharply. Brussels is implementing its AI Act through graduated enforcement. London has so far preferred a lighter-touch sectoral approach, with individual regulators managing AI risk within their own domains. The Lords amendment, if adopted by the Commons, would mark a significant departure from that framework — inserting a centralised override at the top of the architecture.

The government has not yet indicated whether it will accept, reject, or substantially rewrite the provision before the bill returns to the lower house.