UK government rejects calls for an AI “kill switch”

The UK Government has rejected calls for an emergency AI “kill switch”, saying it cannot simply turn AI off. The Cabinet Office told the BBC that blocking access to a model or shutting down specific infrastructure would not prevent the same model from being developed and hosted elsewhere.

The position follows calls from MPs for a statutory trigger that could force a shutdown during an emergency. The government said AI companies also have responsibility for building products with appropriate safeguards.

Former OpenAI researcher Danial Kokotajlo said a legal or technical kill switch controlled by one country would not be feasible. He suggested that companies working together could be one possible solution, while warning that switching off domestic access to a model might do little to protect people from systems developed elsewhere.

The latest debate followed a September 1 discussion in the House of Lords and an open letter warning of a “limited window to strengthen cyber defenses”. More than 100 companies, including OpenAI, Anthropic, Google, Microsoft, AWS and Adobe, urged governments, cybersecurity companies and frontier AI developers to cooperate.

Lord Clement-Jones said security services and regulators currently lack a specific statutory mechanism to compel a physical or digital shutdown. The source reports that companies and lawmakers may need to collaborate more effectively to reduce future risks.