Ted Cruz is targeting Sept. 23 for a Senate Commerce Committee markup of the Klobuchar-Cruz-Thune AI bill, according to three people familiar with the plans reported by Axios. To hold that vote, the chairman must notice the markup by Wednesday, Sept. 17. That short calendar is doing a lot of policy work.

The draft creates a federal “duty of care” that exposes frontier labs to liability for catastrophic risks and, per Reuters sourcing cited by The Next Web, would let Washington block the release of models judged unsafe, with a federal court challenge available. The clause quietly reshaping compliance for everyone downstream is the preemption language: two Senate aides told Reuters, per The Next Web, that it would bar states from enforcing their own laws on certain frontier model risks, including AI-assisted bio and nuclear weapons design. That’s the state patchwork Cruz’s committee is moving to displace, including California SB 53 and the New York RAISE Act.

The politics remain unsettled. Ranking Democrat Maria Cantwell is holding out over who runs safety tests. Majority Leader John Thune told reporters Monday the bill needs a “light touch” that avoids stifling innovation while mitigating “more consequential threats,” and confirmed weekend talks with Klobuchar on a revised draft.

Running parallel: Anthropic, OpenAI, and Google have discussed forming an industry safety body, the Washington Post reported, a familiar move whenever Congress looks like it might actually legislate. The 1996 Telecommunications Act cleared a similar state-federal thicket the same way, by preemption bundled with a duty framework the incumbents helped shape.

For owner-operators already pricing in California’s newly required independent AI auditors, the practical read is to watch Sept. 23 vote and Wednesday’s notice deadline before committing more spend to state-specific compliance work that could be overriding the state compliance patchwork small businesses have been building against within a week.

Sources

Sources