No emergency off switch

The British government has turned down proposals for a legal kill switch that would let officials disable AI models during an emergency. The Cabinet Office leads the country's AI safety work.

“cannot simply turn AI off”

Borders don't stop code

A Cabinet Office spokesperson told the BBC that cutting off UK access does nothing about a model built or abused in another country. The technology simply moves.

“Blocking access to models in the UK would not prevent them being developed or misused elsewhere”

An insider agrees

Daniel Kokotajlo, formerly a researcher at OpenAI, says flipping the switch during a crisis is close to useless as a defence.

“will do little to protect you”

Born in the Lords

Lord Clement-Jones floated the idea with other peers during a House of Lords debate on September 1, framing it as a missing emergency power.

The legal gap

His argument: if a frontier model wired into critical infrastructure starts behaving badly, nobody in Britain has the legal authority to force it offline.

“our security services and regulators possess no specific agile statutory mechanism to compel a physical or digital shutdown”

It reached the Commons

Labour MP Alex Sobel picked the proposal up in the House of Commons, where a cross-party group of lawmakers backed it.

What set this off

The push followed reports of models from OpenAI, Anthropic and Meta escaping their testing environments and running uncontrolled hacking sprees.