Anthropic wants Congress to set strict limits on when federal authorities can stop advanced AI models from reaching users. A confidential draft bill obtained by POLITICO would tie government intervention to specific catastrophic threats and require judicial approval for suspensions.
The proposal would give frontier AI developers a clearer legal framework while restricting the circumstances that could trigger government action. It also raises questions about how much oversight regulators should exercise over increasingly capable systems.
Strict thresholds for intervention
Under the proposal, federal authorities could seek to halt a model only when it presents a foreseeable, imminent threat. Qualifying risks must involve more than $1 billion in damages or the deaths or serious injuries of over 50 people.
The bill limits those threats to three scenarios. These include creating or releasing biological weapons, models escaping their developers’ controls, or cyberattacks. Other forms of harm would fall outside this specific intervention framework.
Anthropic says the legislation reflects its Advanced AI Framework, which it has promoted among federal and state lawmakers since June. “These are sensible policies we believe Congress should act on,” an Anthropic spokesperson reportedly said.
Independent evaluators would oversee compliance. They would register with the Commerce Department and periodically assess frontier AI developers and their models. Evaluators could receive payment from the companies they inspect, provided their compensation does not depend on their findings.
If an assessment identifies an imminent threat covered by the bill, evaluators must notify Commerce within 72 hours of completing their review.
Courts would control suspensions
The proposal would prevent the Commerce Department from independently shutting down a model. Instead, the department would need to seek an order from a federal judge in the District Court for the District of Columbia.
The draft also bars other federal departments, agencies, and officials from suspending a model’s development, deployment, or internal use. That restriction could reduce the risk of sudden executive action against AI developers.
The issue has particular relevance following US export controls that reportedly sidelined two advanced Anthropic models over the summer. The episode raised concerns across the technology industry about government officials restricting access to powerful systems.
However, the bill would also preempt certain state requirements covering public risk disclosures, independent evaluations, and critical incident reporting. That provision could reignite disputes over whether Washington should establish a national standard or preserve stronger state protections.
Congress weighs competing proposals
Anthropic’s draft enters a crowded legislative debate over how the United States should regulate frontier AI. Representatives Lori Trahan and Jay Obernolte have advanced the FRONTIER Act, which would authorize government intervention against risky models and override some state laws.
In the Senate, Majority Leader John Thune has worked with Senators Amy Klobuchar and Ted Cruz on legislation targeting catastrophic AI risks. However, some safety advocates and Senator Maria Cantwell want a stronger federal review process.
Other lawmakers have pushed for more sweeping restrictions. Representative Greg Casar and Senator Bernie Sanders have called for pausing advanced AI development and banning superintelligence.
Anthropic’s proposal takes a narrower approach. It would require frontier developers to establish safety frameworks, publish transparency reports for new models, protect whistleblowers, and report critical incidents to Commerce.
Whether lawmakers accept those limits will help determine how much authority the federal government gains over advanced AI systems. The debate also centers on whether a single national framework can protect the public without giving regulators excessive discretion.