Leaked Anthropic Federal AI Bill Draft Sparks Regulatory Capture Backlash

A discussion draft of a federal AI safety bill authored by Anthropic leaked over a weekend in mid-October. AJ Cimino, Anthropic's head of federal affairs, later confirmed the draft translates the company's "Advanced AI Framework" proposal into legislative language, and that it had been shared with a small number of congressional offices beforehand.

Confirmed

• Politico reported that Anthropic was circulating the discussion draft among lawmakers; Anthropic (via AJ Cimino) confirmed its existence and origin.
• Critics say the draft is notably weaker than the FRONTIER Act from Rep. Obernolte and Sen. Trahan; Transformer News argued it could even make it "legally harder than it is now to block the release of dangerous models."
• Law professor Stephen L. Casper called the draft "pretty bad" after reading it, describing it as a textbook case of regulatory capture that dilutes key provisions of existing legislative drafts, and pointed out the inconsistency with Dario Amodei's safety statements made just weeks earlier.
• Brad Carson — who heads Public First, an organization backed by Anthropic and Amodei — publicly criticized the framework, arguing that companies cannot set their own safety standards, though he first acknowledged some of Anthropic's positions.
• OpenAI policy chief Chris Lehane seized the moment to publish a piece urging Congress to back the FRONTIER Act.
• Anthropic researcher Ethan Perez, resharing the debate, said he disagreed with the draft's language on federal preemption, calling it insufficiently precise.
• Safety researcher David Manheim took a softer stance; one of his concerns is that legislative lobbying may move slower than model iterations.

Why it matters

Anthropic has long projected a "safety first" image, yet the leaked draft suggests the federal legislation it is pushing could actually weaken existing safety provisions — fueling doubts in the safety community about its true stance and corporate self-interest. Even the head of a think tank it funds came out against companies setting their own standards, highlighting the contested boundaries of "corporate-drafted legislation" in AI governance and handing rivals like OpenAI a policy attack line.

2026-10-19 ~ 2026-10-20 · 12 related posts

Full story(2 episodes)→

• Episode 1: Anthropic's secretly pushed AI safety bill draws fire from White House and safety researchers(2026-10-17, 29 posts)
• Episode 2: Leaked Anthropic Federal AI Bill Draft Sparks Regulatory Capture Backlash(2026-10-19, 12 posts)

• [source] Anthropic confirms draft federal AI safety bill translates its Advanced AI Framework into legislative language — edwinarbus · 2026-10-19
• Anthropic researcher flags imprecise preemption language in AI policy draft — EthanJPerez · 2026-10-19
• Anthropic privately pitching Congress a bill: government can block an SI model only above $1B damage or 50 deaths — AIFlow_ML · 2026-10-19
• Anthropic proposes blocking AI models only over $1B damage or 50+ casualties — fsharpman · 2026-10-19
• OpenAI's Chris Lehane pushes Congress to act as Anthropic leak roils AI safety world — ShakeelHashim · 2026-10-20
• [source] Leaked Anthropic draft AI bill panned as weaker than FRONTIER Act, 'safety-first' pitch rings hollow — ShakeelHashim · 2026-10-20
• [source] Law professor slams leaked Anthropic draft AI bill as textbook regulatory capture — StephenLCasper · 2026-10-20
• Anthropic's leaked federal AI bill draft sparks regulatory capture criticism — davidmanheim · 2026-10-20
• Anthropic: block AI models only if they threaten $1B damage or 50+ deaths — Ok-Conversation3213 · 2026-10-20
• Brad Carson of Anthropic-backed Public First criticizes Anthropic's AI framework bill — ShakeelHashim · 2026-10-20
• Anthropic's draft federal AI safety bill falls short of binding standards, watchdog ARI says — wfithian · 2026-10-20
• Nathan Calvin's long post asks one of the biggest questions: what is Anthropic for? — ShakeelHashim · 2026-10-20