When will AI biosecurity safeguards become U.S. law?

The core question is not whether AI biosecurity safeguards are being discussed, but whether a U.S. law specifically requiring them will be enacted before the deadline. On that front, the evidence is encouraging for eventual action but weak for near-term passage: multiple bills have been introduced, there is visible bipartisan and state-level pressure, and some industry actors appear willing to endorse narrower safeguards. Even so, the federal system still shows the usual bottlenecks of committee negotiation, jurisdictional conflicts, and uncertainty over what specific obligations would survive the legislative process. With only a short runway remaining, the default expectation should be that many proposals remain proposals rather than becoming law.

Historically, even issues with broad elite concern often fail to clear Congress quickly, especially when the policy target is technically complex and cross-cuts national security, health, commerce, and AI governance. A biosecurity-focused AI statute would likely need agreement on definitions, scope, enforcement, and who bears compliance costs, and those details can delay passage even when the headline goal is widely supported. The fact that the current federal framework is still mostly executive-branch driven, voluntary, or based on existing authorities suggests that Congress has not yet converged on a durable legislative solution.

Compared with the market price, I see the 27% yes probability as somewhat optimistic. That price appears to be giving substantial credit to the existence of active bills and rising concern, but introduced legislation and public pressure are not the same as enactment. I would expect the market to be overvaluing the probability that one of the current initiatives clears both chambers and gets signed before January 1, 2027. A lower single-digit-to-teens chance seems more consistent with the combination of limited time, legislative inertia, and the absence of any enacted law so far.