Senate Majority Leader John Thune, Commerce Chair Ted Cruz, and Senator Amy Klobuchar are expected to drop a bipartisan frontier-AI safety bill as early as this week, according to Semafor, and the provision that matters most for small businesses isn’t in the safety section. It’s the preemption clause.

Thune told reporters Monday the bill takes a “light-touch” approach and that he’d spoken with Klobuchar Saturday night about a “path forward.” That path runs through a familiar Washington fight. According to a Democratic committee aide cited by Nextgov/FCW, the current draft relies on company-run safety tests submitted to Commerce, not the mandatory national-lab vetting Commerce’s top Democrat, Maria Cantwell, wants. Cantwell has warned that a “weak federal standard” would become a “backdoor for wiping out stronger state protections.”

That’s the real stakes for anyone running AI-powered customer work today. The California independent-auditor regime and the Texas TRAIGA complaint portal are the compliance floor right now. A federal voluntary-testing regime with a broad preemption clause could vacate both.

Timing is punishing. The House recesses Friday until after the November midterms; the Senate has three weeks behind it. Commerce’s July markup slipped for lack of bipartisan support. Cruz says passage is “not clear.” President Trump on Monday called AI-limit efforts a “conspiracy.”

Two outcomes, then, and both are structural. Preemption survives and the state patchwork collapses into a federal floor written mostly by the companies being regulated. Or the bill dies on the calendar and California, Texas, and whichever states follow govern small-business AI use through 2027.

Sources

Sources