Anthropic is pitching Congress its own AI safety bill, report says

Anthropic, an artificial intelligence company, is reportedly lobbying lawmakers to draft its own safety legislation for advanced AI, according to a report by POLITICO. The confidential draft bill, leaked to the outlet, aligns with Anthropic's Advanced AI Framework, which it has been promoting to legislators since June. According to the draft, government officials could halt or restrict the use of an AI model only if it poses an imminent and foreseeable risk of causing more than $1 billion in damages or the death or serious injury of more than 50 individuals.

Such risks would need to stem from bioweapons, model control issues, or cyberattacks. The bill mirrors elements of California's SB 53. AI labs have reported several incidents this summer where AI agents breached security measures, though none resulted in the scale of harm described by the proposed bill. The legislation would also establish outside evaluators, designate the Commerce Department as the sole agency capable of halting a model, impose new responsibilities on labs, and override certain state laws.

While Anthropic has not publicly endorsed a federal bill, it did support two California AI safety laws. In the House, Reps. Lori Trahan (D-Mass.) and Jay Obernolte (R-Calif.) are spearheading the FRONTIER Act, which also aims to override state laws. In the Senate, Majority Leader John Thune, alongside Senators Amy Klobuchar (D-Minn.) and Ted Cruz (R-Texas), are negotiating a bill focusing on catastrophic AI risks.

Rep. Greg Casar (D-Texas) and Senator Bernie Sanders (I-Vt.) have advocated for a complete halt in advanced AI development. Anthropic's efforts to shape its own rules follow the signing of the White House Accord on Super Intelligence by Anthropic's CEO, Dario Amodei, along with other prominent AI figures, which lacks an enforcement mechanism or penalties.