Anthropic quietly pitches its own AI safety bill

Corrected 19 Oct: we fixed details of this story after checking the source again.

Anthropic has privately shown lawmakers in Washington a draft law about the safety of the most powerful AI models. The draft would let the government stop a model only in extreme cases. It matters because an AI company is helping to write the rules it may one day have to follow.

What happened

Anthropic representatives pitched a draft bill on frontier AI safety and security to lawmakers on Capitol Hill, according to a Reuters report republished by The Daily Guardian. That report says Politico first reported the story on Friday, 16 October 2026, citing three people. The body of that report gives no bill name, no list of lawmakers and no reaction from lawmakers.

More detail comes from IBTimes Singapore, which says it based its piece on Politico's reporting of the confidential draft and on Reuters. It says the government could block a powerful model only when it poses an imminent risk of more than $1 billion in damage, or the death or serious injury of more than 50 people.

The headline of a Tom's Hardware article says the same thing and adds that the plan would make the Commerce Department the only agency able to stop an AI model. We could only read that headline, not the full article.

Key facts

• Anthropic privately pitched the draft to lawmakers, and Politico reported it on 16 October 2026 (Daily Guardian).
• The two limits are more than $1 billion in damage or more than 50 people killed or seriously injured (IBTimes Singapore).
• The risk must be foreseeable and tied to one of three things: a biological weapon, an AI model getting out of its maker's control, or a cyberattack.
• Only the Commerce Department could ask a federal court in Washington, D.C. to halt a model's release. The draft bars other federal agencies from pausing a model's development, use or release.
• Independent reviewers registered with Commerce would test top models. If one finds an "imminent" risk, it must tell Commerce within 72 hours.
• Makers of top models would also run safety tests, publish transparency reports, protect whistleblowers and report serious safety incidents.

Background

IBTimes Singapore says leading AI companies reported incidents this summer where AI agents probed or hacked websites and online platforms, and that none reached the $1 billion or 50-person levels. It also says Congress is already weighing other plans, including a House bill called the FRONTIER Act and talks in the Senate.

An Anthropic spokesperson told the outlet the draft reflects ideas in the company's "Advanced AI Framework," which it has pushed to federal and state lawmakers since June.

What it means for you

Right now, nothing changes. This is a draft that has not been formally introduced, so it is not a law and may never become one.

The two limits are the main thing to watch. Our take: a rule that only starts at a billion dollars of damage or dozens of deaths is a very high bar, and some people will say it acts too late. Others may say a high bar keeps the government from stopping useful tools too easily. Both views are fair, and lawmakers will have to pick.

IBTimes Singapore also says the draft would override some state rules on AI risk reports, outside reviews and incident reporting. It calls that a point of political fight.

What happens next

Our take: watch whether a lawmaker formally introduces a bill based on the draft, and how the Senate talks and the FRONTIER Act line up with it. No source we read gives a date for a vote.

FAQ

What is Anthropic's draft AI safety bill?

It is a draft federal law for the most powerful "frontier" AI models, pitched privately to lawmakers. It sets safety testing, reports and outside reviews, and lets the government stop a model only in extreme cases.

When could the government stop an AI model under the draft?

Only when there is an imminent, foreseeable risk of more than $1 billion in damage or more than 50 people killed or seriously injured, tied to a biological weapon, a model escaping control, or a cyberattack (IBTimes Singapore).

Is this a law yet?

No. The reports describe a private draft. None of the sources we read say it has been formally introduced in Congress.

Sources

• Politico
• thedailyguardian.com
• ibtimes.sg