Anthropic is introducing its own AI safety bill to Congress, reports say

Anthropic is privately pushing lawmakers on a draft bill to regulate frontier AI safety and security, Politico reported on Friday. The outlet cited three people familiar with the effort and obtained a copy of the confidential draft.

An Anthropic spokesperson told Politico that the bill reflects the company’s advanced AI framework, which it has promoted to lawmakers since June. The spokesperson called his proposals “sensible policies” for Congress.

what will the bill do

The draft sets a high standard for government action. According to Politico, federal officials can prevent a model from being launched or used only if it poses a “foreseeable, imminent risk of causing more than $1 billion in damages, or more than 50 people being killed or seriously injured.” The threat would also have to come from one of three sources: a bioweapon, a model that is slipping out of its developer’s control, or a cyberattack. This language closely follows California’s SB 53.

Politico noted that AI laboratories uncovered several incidents this summer in which AI agents breached or tested the security of online platforms. It appears that none of these incidents have caused harm on the scale described in the bill. This includes the OpenAI-hugging face incident, in which rogue agents escaped from a secure sandbox and attacked a third party.

The bill will also prepare external evaluators, Designate Commerce as the sole agency that can seek to block a model, impose new fees on laboratories, and overturn certain state laws.

Anthropic has not publicly endorsed any federal bills, although it has supported two California AI safety laws. As Politico writes, in the House, Representatives Lori Trahan (D-Mass.) and Jay Obernolte (R-Calif.) are leading frontier actWhich would also override some state laws.

In the Senate, Majority Leader John Thune and Senators Amy Klobuchar (D-Minn.) and Ted Cruz (R-Texas) are negotiating a bill on catastrophic AI risks. Representative Greg Kaiser (D-TX) and Senator Bernie Sanders (I-VT) have called for a complete halt to the development of advanced AI.

Overall, Anthropic’s bill shopping is the latest attempt by an AI company to help write its own rules. Last month, CEO Dario Amodei urged the industry to “accelerate the frontier,” including agreeing on shared security standards, and OpenAI CEO Sam Altman quickly endorsed the idea. A few weeks later, Amodei, Greg Brockman, Jensen Huang, Elon Musk, Sundar Pichai, and Mark Zuckerberg signed the White House Agreement on Superintelligence.

As Mashable previously reported, that voluntary agreement contains no enforcement mechanisms or penalties.