45 States Wrote AI Laws. The White House Just Told Them to Stop.

On Friday, the White House released a four-page document that could erase two years of state-level AI regulation in one stroke. The framework asks Congress to preempt state AI laws and replace them with a "minimally burdensome national standard."

It landed on the same day that 1,561 AI-related bills sit active across 45 state legislatures.

What the Framework Actually Does

The document itself is thin — four pages, seven broad categories. Kids' safety, copyright, jobs, censorship, and the big one: preemption. It tells Congress to stop states from regulating AI development entirely, calling it an "inherently interstate" issue tied to national security.

States can still enforce general fraud laws and set zoning rules for data centres. But the framework draws a hard line against any state telling AI companies how to build, test, or deploy their models. It would also shield developers from liability for how third parties use their tools.

This isn't new. In December 2025, Trump signed an executive order directing the DOJ to create an AI Litigation Task Force — a unit whose job is to sue states over AI laws in federal court. That task force has been active since January 10. The Commerce Department was given 90 days to compile a list of "onerous" state laws. States that make the list risk losing federal broadband funding.

What States Already Built

Colorado's AI Act took effect February 1, 2026 — the first comprehensive state AI law in the country. New York's RAISE Act requires safety protocols for large AI systems. California's SB-53 mandates that AI companies document and publish their safety practices.

These aren't theoretical. Companies are already doing compliance work. Impact assessments, monitoring logs, disclosure requirements. Colorado's attorney general can enforce violations right now.

The federal framework would override all of it.

The Person Writing the Rules

White House AI czar David Sacks is a venture capitalist whose firm, Craft Ventures, invested $22 million in an AI company targeting federal contracts — seven months after he promised to divest. Senator Elizabeth Warren launched an ethics investigation. The SF Standard reported that Silicon Valley "rolled its eyes" at the conflict-of-interest allegations.

Sacks' framework reads exactly as the AI industry wanted it to. NetChoice, a tech industry group, praised it for showing what "it will take to win the future." The Alliance for Secure AI called it doing "the bidding of Big Tech."

The Timing Problem

The EU AI Act's remaining provisions take full effect August 2, 2026 — four months from now. Fines up to €35 million or 7% of global revenue. Binding, enforceable, already being implemented.

The US framework is a request to Congress. Four pages. No enforcement mechanism. No new regulatory body. No timeline for legislation.

Two approaches to the same technology, running at completely different speeds. Europe's betting that rules attract trust. Washington's betting that rules slow you down. Both can't be right — but every AI company operating globally has to comply with both.

The 1,561 state bills weren't chaos. They were 45 laboratories trying to solve a problem the federal government refused to touch. Now Washington wants to stop the experiments before seeing the results.