Anthropic Pitches Draft AI Regulation Bill to US Congress

Anthropic has quietly pitched a draft AI regulation bill to US lawmakers that establishes strict thresholds for government intervention and seeks to override fragmented state-level laws. The proposal dictates that the government can only block an AI model if it poses severe, imminent risks such as causing over $1 billion in economic damage or more than 50 deaths and severe injuries. This initiative highlights how leading AI labs are actively trying to shape federal legislation to prevent a complex patchwork of state-level regulations while securing high intervention thresholds for their technologies. By preempting state laws, the bill could streamline compliance for major tech companies operating across the United States. The proposed bill restricts government shutdowns to specific catastrophic scenarios like biological weapon proliferation, loss of model control, or severe cyberattacks. Additionally, frontier AI developers would be required to establish safety frameworks, publish transparency reports, enforce whistleblower protections, and report major incidents to the US Department of Commerce.