Just passed on Thursday, May 22nd, by the United States House in a razor-thin 215–214 vote, the so-called “One, Big, Beautiful Bill” (H.R. 1) doesn’t just touch tax reform and border policy. Buried deep inside is a 10-year moratorium on any state or local law regulating artificial intelligence.
“No State or political subdivision thereof may enforce any law or regulation regulating artificial intelligence models, artificial intelligence systems, or automated decision systems…”
— Section 43201(c), H.R. 1
That one line—tucked into a section about federal IT modernization—could instantly wipe out dozens of state-level AI laws. Think: hiring bias protections, deepfake bans, algorithmic transparency rules. Gone. Zip. Kaput.
Now we just wait to see what the Senate does with it.
A ban without a backup plan
There’s no federal framework to replace what states have been building. Just a full stop. No standards. No oversight. Just “don’t do anything for a decade.”
There’s a narrow exception for laws that carry criminal penalties, but that’s it. Nothing for civil enforcement. Nothing for consumer protection. Nothing for the privacy harms and discrimination that states and advocacy groups have actually been showing up for.
Let’s be real: the people pushing back on AI harms—privacy advocates, civil rights organizations, local lawmakers—aren’t doing it because it’s trendy. They’re doing it because the federal government hasn’t. This bill wouldn’t just block overreach—it would block the people trying to keep the tech industry in check.
Preemption without the clause
What makes this all even stranger? It’s not structured like real preemption. Normally, when Congress wants to take something out of the states’ hands, they include a formal preemption clause—something like:
“This Act shall supersede any and all State laws to the extent that such laws regulate [subject matter].”
That’s how you usually invoke the Supremacy Clause—clearly, directly, and with the legal weight to back it up.
H.R. 1 doesn’t do that. It just bans enforcement of state laws for 10 years, with no underlying federal rules, no standards, no real justification beyond “we’ll figure it out later.” It’s preemption without saying it’s preemption.
And it’s not even in a standalone tech bill. It’s stuck inside a budget reconciliation bill—a tool meant for taxes and federal spending—not sweeping national tech policy.
The Byrd Rule wildcard
Under the Senate’s Byrd Rule, anything “extraneous” to the federal budget can get struck from a reconciliation bill. This AI ban lives in a clause about modernizing government IT systems, but whether that’s enough budget tie-in to survive? That’s a real question.
If the Senate parliamentarian rules it out of bounds, it would take 60 votes to override, and that’s a high bar.
What happens next?
Now the bill heads to the Senate. Even if the larger package moves, this AI moratorium might not make it—not because of the politics, but because of the process. And honestly, that might be the only thing stopping it.
A quiet but consequential shift
Supporters of the bill say this is all about avoiding a confusing patchwork of state laws. And sure, state rules don’t always align. But a decade-long freeze, with nothing in place to fill the gap, isn’t solving that. It’s erasing it.
This isn’t comprehensive AI regulation. It’s a placeholder. A federal AI pause—without standards, without structure, and without a proper preemption clause.
We don’t need quiet preemption. We need clear rules, public input, and actual governance.