Anthropic has privately approached U.S. lawmakers with a draft bill designed to regulate advanced artificial intelligence models.
This was first reported on Friday by Politico, citing three sources familiar with the matter.
Representatives from the AI developer presented the draft legislation to lawmakers on Capitol Hill, outlining a regulatory framework intended to ensure the safety and security of advanced AI models.
Under the proposal, government intervention would be permitted only if an AI model poses a foreseeable and imminent risk of causing damages exceeding $1 billion, or resulting in the death or serious bodily harm of more than 50 people.
Furthermore, such risks must be directly linked to the creation or proliferation of biological weapons, cyberattacks, or scenarios in which an AI model circumvents the safety mechanisms established by its developers.
The bill proposes the creation of a system of independent expert evaluators, registered with the U.S. Department of Commerce, who would conduct periodic audits of AI developers and their advanced models.
AI development companies would be permitted to pay for the services of these evaluators, provided that the compensation is not contingent upon the outcome of the assessment.
If an evaluator identifies an imminent risk that meets the criteria specified in the bill, they would be required to notify the Department of Commerce within 72 hours of completing the assessment.
Following such a notification, the Department could petition a court to halt the release or use of the model, should the identified risks necessitate intervention.
At the same time, the proposed legislation would prohibit other federal agencies, departments, or officials from independently suspending the development, deployment, or internal use of AI models. The authority to issue such a suspension order will rest exclusively with a federal judge from the U.S. District Court for the District of Columbia.
Anthropic and the U.S. Department of Commerce did not respond to requests for comment.
This initiative reflects growing efforts to establish federal oversight of increasingly powerful AI systems, while simultaneously limiting the grounds upon which regulators can restrict their development or commercial deployment.