Anthropic pitches AI safety bill to US lawmakers, proposes restrictions on dangerous models

Anthropic has proposed a draft law to regulate the safety and security of advanced artificial intelligence systems, giving the US government a mechanism to seek restrictions on models that pose an imminent threat to human life or property, Bloomberg reported.

The proposal, which the company has presented to lawmakers on Capitol Hill, would allow authorities to seek a halt to an AI model's release or use if it presents a foreseeable and immediate risk of killing or seriously injuring more than 50 people, or causing over $1 billion in damage, according to Politico, which first reported the draft legislation.

The threshold would apply to risks arising from the creation or release of biological weapons, an AI system escaping its developer's control, or cyberattacks, the report said.

An Anthropic spokesperson told Bloomberg the draft legislation reflects recommendations in the company's Advanced AI Framework, which it has been advocating to federal and state lawmakers since June.

Bill proposes independent evaluations of AI models

The proposed legislation would require developers of frontier AI models to conduct safety tests and make their findings public. Companies would also have to permit independent assessments of their models at regular intervals, according to the spokesperson.

Independent evaluators would register with the US Department of Commerce and periodically assess AI developers and their most advanced models. The draft would allow companies to pay for these assessments, provided compensation was not linked to the outcome of an evaluation, Politico reported.

If evaluators identify an imminent risk covered by the legislation, they would have to notify the Commerce Department within 72 hours of completing their assessment, the report said.

The department could then ask a federal judge in the District of Columbia to suspend the model. The draft reportedly limits this suspension mechanism to the specified process, barring other federal departments, agencies and officials from independently stopping a model's development, deployment or internal use.

Anthropic said the proposed framework would give the federal government the power to prevent or discourage dangerous deployments of the most powerful AI systems. The company called on Congress to act on the measures.

The proposal comes amid growing concern over the risks posed by increasingly capable AI systems. Recent breaches and hacks involving AI models have intensified calls for safeguards, although the administration of US President Donald Trump has so far rejected demands for greater oversight.

Follow Storyboard18 on Google for the latest and breaking digital news and industry updates, along with in-depth coverage of digital media and gaming news. Stay informed with the latest perspectives only on Storyboard18.