# Anthropic is pitching lawmakers a draft federal AI safety bill

> Politico reported, and Reuters picked up, that Anthropic pitched a draft bill to block a model only above a stated harm line.

Published 2026-10-16T17:20:00-07:00 · News · Canonical: https://news.littlewolflabs.com/posts/2026-10-16-anthropic-federal-safety-bill

**Reported.** On Oct 16, [Politico reported](https://www.politico.com/news/2026/10/16/anthropic-ai-safety-01124654) that Anthropic representatives have quietly pitched Capitol Hill lawmakers a draft bill on frontier AI safety and security. [Reuters](https://www.centraloregondaily.com/news/consumer/anthropic-quietly-pitches-its-own-ai-safety-bill-to-lawmakers-politico-reports/article_dc499abb-13c9-5242-84db-81b14bc875f1.html) carried the same account the same day, citing Politico and three people with knowledge of the efforts. The copy linked here is the Reuters wire as Central Oregon Daily published it. This is not a bill that has been introduced, and Anthropic has not posted the draft.



Reuters says the draft would let the government block a model's release or use only if it poses a foreseeable, imminent risk of more than $1 billion in damages, or of killing or seriously injuring more than 50 people. Politico, as relayed by Reuters, says that risk has to come from a biological weapon, the model evading its developer's controls, or a cyberattack.

An Anthropic spokesperson told Reuters the bill represents proposals in Anthropic's [Advanced AI Framework](https://www.anthropic.com/policy-on-the-ai-exponential), released earlier this year, and that it would give the government the ability to block or deter dangerous deployments of the most powerful models. The spokesperson also said the bill would require developers to test frontier models, be transparent about the findings, and have them independently evaluated. The [framework PDF](https://www-cdn.anthropic.com/files/4zrzovbb/website/0a58d567024a8b448ff15158ebc3625328dfcc1f.pdf) is Anthropic's June 2026 proposal. On its policy page, Anthropic says frontier developers should test models, publish what they find, use independent evaluators, and that government should be able to block or deter dangerous deployments, with safeguards against overbroad power. The page also says Congress should not preempt state law unless the federal law is at least as strong as that framework.



Reuters, citing Politico, adds these reported details. Independent evaluators would register with the Commerce Department and periodically evaluate the lab and its frontier models. Labs could pay those evaluators if the payment is not tied to the outcome. If evaluators find an "imminent enumerated risk," they must tell Commerce within 72 hours of finishing the assessment. Commerce could ask a judge to halt a model. The draft, as reported, would bar any other federal department, agency, or officer from suspending development, deployment, or internal use. A federal judge in the District Court for the District of Columbia could order the suspension. Reuters says the Commerce Department did not immediately respond to a request for comment.

## What it costs

The reported draft does not publish a fee, a fine schedule, or a price for anyone using a model. Nothing on the pages above is a product you can buy.

## Should you switch

There is nothing to install or cancel. The draft is reported, not enacted. If you follow US rules for frontier models, the thing to watch is whether a member of Congress files the text, and how it compares with the framework Anthropic already published.