Jack Clark, one of the seven founders of Anthropic, said on the BBC that a way to shut off AI software completely if it becomes too dangerous is something society might want to eventually pass rules around. He added that while most labs already have mechanisms to pull the plug, lawmakers may need to enforce a mandatory kill switch that can be verified by a third party.
Clark’s comments come as AI’s rapid development and fears over its risks have drawn attention from executives and staff at major firms. Anthropic chief executive Dario Amodei has called for a slower pace of AI development and closer monitoring, while others in the industry have questioned the motivations behind such calls.
In the United States, lawmakers have drafted the Kill Switch Act, which would require companies to have a way to shut down problematic AI tools and would give certain agencies the power to demand a tool be turned off or limited. The act remains under debate, and President Donald Trump has dismissed concerns about AI as a hoax.
In the United Kingdom, the government recently rejected the idea of creating a kill switch, saying it would not prevent misuse elsewhere. Anthropic, valued at $965 bn, is preparing for an initial public offering, while OpenAI, valued at $852 bn, has postponed its IPO due to the safety debate.
Clark said, "We are rolling dice with immense risks," and that allowing AI to continue as a totally unregulated industry was a bad idea. He emphasized that any action to rein in AI development should not sacrifice commercial advantage.
As the conversation around AI safety intensifies, stakeholders will watch how governments and regulators respond to calls for mandatory kill switches and other safeguards.
