'kill switch' may need to be mandatory, Anthropic co-founder says

Jack Clark, co-founder of Anthropic, said an AI “kill switch” that a third party can check may need to become mandatory for companies. Clark, one of seven Anthropic founders, told the BBC that most labs have different ways to shut down software if it becomes too dangerous, but policymakers may need to enforce requirements and verification. His comments come amid a broader debate in which more AI executives and experts have raised the possibility that AI could harm or kill humans. Clark pointed to the need for a “larger policy conversation” around specifics, including whether such a kill switch is verifiable. Anthropic’s Dario Amodei previously called for slowing AI development and closer monitoring, while Clark said regulation should not sacrifice commercial advantage.