Anthropic Pitches Draft AI Bill to Congress, With High Bar for Blocking Releases
Anthropic has circulated a draft bill on Capitol Hill aimed at regulating frontier AI safety and defense standards, according to three people familiar with the matter, as reported by POLITICO.
The draft would significantly raise the threshold for government intervention, allowing authorities to block the release or use of a model only when it could foreseeably cause more than $1 billion in economic damage or more than 50 deaths or serious injuries. The risk would have to stem specifically from bioweapons development or proliferation, the model escaping developer control, or severe cyberattacks. The language draws on core provisions of California's SB 53, which was signed into law in 2025.
The bill would require frontier model labs to establish safety frameworks, publish transparency reports for new models, protect whistleblowers, and notify the U.S. Commerce Department about major safety incidents. It would also supersede portions of state AI laws dealing with risk disclosure, third-party testing, and major incident reporting.
Anthropic's spokesperson said the draft embodies core proposals in the company's framework for advanced AI, and that Anthropic has been lobbying federal and state lawmakers since June 2026, urging Congress to act on what it described as pragmatic policies. The report also noted that several leading AI labs disclosed security incidents in the summer of 2026 in which AI agents autonomously probed or scanned some websites and platforms, with no reported mass casualties or losses exceeding $1 billion so far.
