Anthropic representatives have quietly hand-delivered a draft regulatory framework to Capitol Hill lawmakers, according to a Friday report by Politico. The pitch centers on a narrow, high-stakes trigger for government intervention: the ability to block a model’s release or use only if it poses a foreseeable, imminent risk of more than $1 billion in damages, or of killing or seriously injuring more than 50 people.
The mechanism is specific. The risk must stem from the creation or release of a biological weapon, the model evading its developer’s controls, or a cyberattack. Under this structure, independent evaluators registered with the Commerce Department would periodically assess the lab and its frontier models. These evaluators could be paid by the labs, provided the compensation is not tied to the evaluation's outcome. If they find an “imminent enumerated risk,” they must report it to Commerce within 72 hours of completing their assessment.
The bill draws a hard line on who holds the power to pull the plug. While the Commerce Department could request a judge to halt a model after determining risk, the language explicitly prohibits any other US government department, agency, or officer from suspending development, deployment, or internal use. A federal judge for the District Court for the District of Columbia could order the suspension, but the executive branch’s broader authority remains constrained to this single pathway.
Neither Anthropic nor the Department of Commerce immediately responded to Reuters requests for comment. The lobbying effort signals a shift from open-ended safety debates to codified, quantitative thresholds for intervention. Lawmakers now face the decision of whether to adopt these specific financial and casualty benchmarks as the legal standard for regulating frontier AI.