00:00Anthropic is proposing a new federal law that could allow the government to block an AI model
00:05if it poses an imminent risk of killing more than 50 people or causing over $1 billion in property damage.
00:12The draft legislation focuses on extreme threats, including biological weapons,
00:17major cyber attacks, and AI systems that escape their developers' controls.
00:22Independent evaluators would assess the risks,
00:25with the Commerce Department seeking a court order to halt a dangerous release.
00:30The proposal could mark a significant step toward federal AI oversight,
00:34but it also raises questions about who should decide when the technology has gone too far.
