Why Anthropic’s draft bill sets $1 billion threshold for AI risk - Politico

Anthropic has drafted legislation setting a threshold for when the government could block the release of an artificial intelligence model, and has pitched the draft to lawmakers, Politico reported on October 21, 2026. Under the draft, the government would be permitted to block a model only if it could kill or seriously injure 50 people or more, or cause $1 billion in damage. Those are the two conditions the article states.

Politico reports that the proposed threshold mirrors one used in a California law, but that Anthropic's draft repurposes it: in the draft, the figure is used to decide when a model's release should be blocked. The article does not describe the California law's original function beyond noting the mirroring, and it does not name the lawmakers to whom the draft was pitched. The two numbers Politico foregrounds are 50 — the count of people who would have to be killed or seriously injured — and $1 billion, the damage figure, which the headline uses as the story's framing.

The draft places a company-authored proposal into a legislative process in which, per the article, no bill text has been described beyond the two thresholds. The article presents the threshold as the mechanism that would govern federal blocking authority, and treats the link to California's existing law as the draft's precedent. Politico's account is limited to the threshold and its state-law parallel; it does not report a sponsor, a timeline, or any response from lawmakers or other AI companies.