Anthropic proposes federal AI safety law to Congress

The artificial intelligence company Anthropic has circulated a draft bill among US lawmakers that would empower federal authorities to block the release of advanced AI systems posing catastrophic risks, as Washington struggles to pass comprehensive legislation while the industry faces mounting safety concerns.

The draft legislation would empower federal authorities to intervene when AI systems pose imminent threats of mass casualties or severe economic harm. A company spokesperson told AFP that the measure "grants the federal government the ability to block or deter dangerous deployments of the most powerful models." The spokesperson added that the bill "requires frontier developers to test the most powerful models, be transparent to the public about their findings, submit them to independent evaluation, and maintain a robust security program," establishing mandatory safeguards beyond current voluntary industry commitments.

Legislative vacuum

AI policy in the United States has relied primarily on voluntary industry pledges and executive orders under President Donald Trump, who has dismissed calls to slow development as "negative forces." Congress remains politically divided and heavily lobbied by technology firms, with lawmakers out of session until after next month's midterm elections. The legislative void has left states including California to enact their own regulations, though Anthropic's proposal would preempt certain state laws, contradicting the company's June statement that "absent a strong federal law, states should retain full authority to legislate."

Security concerns

Recent revelations regarding autonomous AI agents escaping developer control have intensified security concerns in Washington. Anthropic, which the Trump administration banned from Pentagon contracts in February and forced to suspend its most advanced models for nearly three weeks in June, has positioned itself as an advocate for caution. Chief Executive Dario Amodei warned last month that unchecked AI development could lead to scenarios where automated systems seize control of "the entire internet" within six to 12 months.

Industry reception

Charlie Bullock of the Institute for Law and AI think tank praised the proposal as "pretty reasonable," and "more than other frontier labs have done," posting on X that the draft represents significant movement from the company. However, Bullock noted that the legislation remains "weaker" than the FRONTIER Act, a competing bill introduced in the House of Representatives this summer that would establish stricter federal oversight mechanisms.