When will a catastrophic AI duty of care become U.S. law?

As of Thu, 19 Nov 2026 10:28:14 UTC, Kalshi traders price Before Jan 1, 2028 (33.0%), Before Jul 1, 2027 (17.0%), Before Jan 1, 2027 (5.0%). The market has $2K in volume and closes on Fri, 07 Jan 2028.

Updated:

Current odds

OutcomeChanceYes priceNo price
Before Jan 1, 202833.0%33.0¢67.0¢
Before Jul 1, 202717.0%17.0¢83.0¢
Before Jan 1, 20275.0%5.0¢95.0¢

Market details

Resolution rules

Each outcome is a separate contract with its own rules. The rules for Before Jan 1, 2028:

If federal legislation imposing an AI catastrophic-risk duty of care has become law after Issuance and before Jan 1, 2028, then the market resolves to Yes. A qualifying federal law must impose an enforceable standard requiring AI developers or deployers to take reasonable care, or equivalent precautions, to prevent or mitigate catastrophic harm from their models or systems. Catastrophic harm refers to large-scale harm such as mass casualties, major disruption of critical infrastructure, or comparably severe economic or environmental damage; the law need not use the exact phrase "duty of care". A law requiring only a framework, audit, report, or shutdown control does NOT count. A framework qualifies only when the law makes the adequacy of the developer's precautions against catastrophic harm an enforceable obligation. Government enforcement or civil liability may enforce the duty. A later effective date does not delay qualification. An example of this is the catastrophic-risk duty-of-care proposal being negotiated by Senators John Thune, Ted Cruz, and Amy Klobuchar.

Sources

Related markets

More markets like this

A price on DOXA is the market-implied probability of an outcome on the venue that lists it. It is what traders are paying, not a forecast made by DOXA, and it is not investment advice. How DOXA shows prices · Terms of Use

DOXA is a liquidity aggregator for prediction markets that compares odds and depth across Polymarket, Kalshi and others and routes each trade for best execution with an AI assistant.