Anthropic's Draft Federal Frontier-AI Bill Leaks; Semafor Reports Safety Advocates Say It Pre-Empts State Laws and Accepts Too Much Risk

Semafor reported on October 20, 2026 that a draft of a federal AI safety proposal Anthropic has been pitching to lawmakers on Capitol Hill leaked “late last week.” A company spokesperson told Semafor the draft is a legislative version of an AI safety framework the company has been pitching for months. Semafor wrote that Anthropic, “long seen as the safety-promoting frontier lab,” found itself “on the defensive,” with one of its top lobbyists posting about tweaks the company plans to make.

On the threshold, the only source here is secondhand. Blood in the Machine describes a bill under which the government could block a model’s release or use only when it poses a foreseeable, imminent risk of more than $1 billion in damages or of killing or seriously injuring more than 50 people, stemming from a biological weapon, the model evading its developer’s controls or a cyberattack. Semafor says AI safety advocates argue the draft “doesn’t go nearly far enough, unfairly pre-empts state laws, and accepts too much possible risk.” OpenAI Chief Global Affairs Officer Chris Lehane wrote that “No one company (including OpenAI) should write the rules alone.” The author of the Blood in the Machine post argues that the most accurate reading of the bill is that it is “a remarkably self-serving one.”

The draft text was not available to this entry, so the threshold comes from that post, not from the text, and the scope of the pre-emption is an advocates’ characterization, not a reading of the text. Related entries: 2026-07-07–openai-anthropic-pre-release-government-control-operational and 2026-07-16–ai-labs-ceos-converge-mandatory-scrutiny-governance-body. The Anthropic details were checked by Mark in a non-Claude verification pass before this entry was written.