Anthropic has drafted a confidential AI safety bill for Congress, proposing federal intervention only when AI systems cause over $1 billion in damages or harm more than 50 people, according to POLITICO.
TL;DR
- Anthropic circulates a draft AI safety bill to Congress, setting high thresholds for federal intervention.
- The bill focuses on three specific risk categories: bioweapons, control evasion, and cyberattacks.
- Anthropic's proposal aims to limit federal agency power and concentrate enforcement in a single court.
What happened
Anthropic has quietly circulated a confidential draft bill on frontier AI safety and security to lawmakers on Capitol Hill, according to POLITICO. The bill proposes federal intervention only when AI systems cause over $1 billion in damages or harm more than 50 people. The draft bill sets a narrow definition of catastrophic risk, focusing on three specific categories: the creation or release of a biological weapon, a model evading its developer's controls, or a cyberattack.
The enforcement mechanism in the draft bill runs through independent evaluators registered with the U.S. Department of Commerce. These evaluators would assess frontier AI companies and their models, reporting any imminent risks to the Commerce Department within 72 hours. The Commerce Department would then petition a federal judge in the U.S. District Court for the District of Columbia to halt the model's release or use if necessary.
Why it matters
Anthropic's proposal marks one of the most concrete attempts yet by a frontier AI lab to turn its own policy preferences into binding federal law. The bill aims to settle the argument over AI oversight on terms that Anthropic can live with, focusing on a narrow definition of catastrophic risk and limiting federal agency power.
The proposal is already drawing attention because it reads less like a wish list and more like a markup-ready statute. It shifts Anthropic from commentator to drafter, signaling the company's intent to shape AI regulation proactively. However, the bill's narrow focus and high thresholds for intervention may face criticism from lawmakers and rival labs seeking broader authority over AI systems.
Key facts
- Anthropic drafted a confidential AI safety bill for Congress, proposing federal intervention when AI systems cause over $1 billion in damages or harm more than 50 people, according to POLITICO.
- The bill focuses on three specific risk categories: bioweapons, control evasion, and cyberattacks.
- Independent evaluators registered with the U.S. Department of Commerce would assess frontier AI companies and their models, reporting imminent risks within 72 hours.
- The Commerce Department would petition a federal judge in the U.S. District Court for the District of Columbia to halt a model's release or use if necessary.
- The bill aims to limit federal agency power, prohibiting other U.S. government departments from independently suspending AI development or deployment.
- Anthropic has been advocating for its Advanced AI Framework, which the bill represents, since June.
- The draft bill is already drawing attention from lawmakers and rival labs, signaling its potential impact on AI regulation.
Context
Anthropic's lobbying push for a uniform federal standard over a patchwork of state rules has been ongoing. The company's leadership, including CEO Dario Amodei, has argued for a uniform federal approach to AI regulation, acknowledging the tension between federal preemption and state urgency. The decision to circulate actual bill text is an attempt to resolve this tension in Anthropic's preferred direction: getting a federal statute in place.
The AI policy fight in 2026 has been defined by the tension between federal preemption and state urgency. States have moved on their own timelines on AI transparency and safety rules, while Congress has debated competing visions for a national standard. Anthropic's proposal aims to address this tension by providing a concrete framework for federal intervention in AI safety.
