Washington's Push to Preempt State AI Laws Heads for a Court Fight

Washington's Push to Preempt State AI Laws Heads for a Court Fight

When President Trump signed Executive Order 14365, "Ensuring a National Policy Framework for Artificial Intelligence," on December 11, 2025, he set the federal government on a collision course with the states over who gets to write the rules for the most consequential technology of the decade. Nearly eight months later, that collision is no longer theoretical. It is unfolding in federal courtrooms in California and Colorado, in memoranda circulating through the Department of Justice, and in the funding formulas that determine which states get billions of dollars in federal grants.

The order directs the administration to build a "minimally burdensome national policy framework for AI" and to clear away what it casts as a thicket of conflicting state rules. It instructs Attorney General Pam Bondi to stand up an "AI Litigation Task Force," tasked with challenging state AI statutes on Commerce Clause, preemption, and other grounds. It also directs the Commerce Department to make states with "onerous" AI laws ineligible for certain broadband (BEAD) funds "to the maximum extent allowed by Federal law," and urges agencies to condition discretionary grants on states agreeing not to enforce conflicting AI rules.

David Sacks, the White House AI and crypto czar, framed the stakes in blunt terms at the Oval Office signing. The proliferation of state bills, he said, had produced a "confusing patchwork of regulation." The administration would not challenge everything, he added, carving out areas like child safety, "but we're going to push back on the most onerous examples of state regulations."

The Litigation Arrives

The task force did not stay on paper for long. In a January 9 memorandum to all Justice Department employees, Bondi formally launched the unit, citing the president's directive to secure U.S. "dominance across many domains." Its stated mission is to contest state laws that "unconstitutionally regulate interstate commerce," are preempted by federal action, or are "otherwise unlawful."

Running in parallel, and now partly merged with the federal effort, is a wave of private litigation led by Elon Musk's xAI. On December 29, 2025, xAI sued California Attorney General Rob Bonta to block Assembly Bill 2013, which requires developers of generative AI systems made available to Californians since January 2022 to publish documentation about their training data. xAI argues the law effects an uncompensated taking of trade secrets under the Fifth Amendment's Takings Clause, compels speech in violation of the First Amendment, and is unconstitutionally vague. The early returns have been mixed for the company: a federal court declined to enjoin the law, finding it likely regulates commercial speech and therefore faces only intermediate scrutiny.

On April 9, 2026, xAI opened a second front, suing Colorado Attorney General Philip Weiser to block the Colorado AI Act (SB 24-205), the nation's first comprehensive algorithmic-discrimination statute. The Justice Department has since joined that case, converting a private grievance into a test of the administration's broader preemption theory.

The Takings Clause argument, in particular, has drawn scrutiny from legal scholars, who note that treating disclosure mandates as compensable takings could reshape the government's power to require transparency far beyond AI. As analysts at Lawfare have observed, the theory is novel enough that its success is far from assured.

The States Dig In

State officials have signaled they will not yield. Bonta warned even before the order was signed that California was prepared to challenge its "potential illegality." In a letter to congressional leaders, he argued that broad preemption "would seriously undermine the federalist system that has always allowed states to respond swiftly and effectively to emerging technologies."

"States — with our flexibility and responsiveness to the market, changing technology, and the needs of our residents — are on the front lines of consumer protection, including when it comes to emerging technology," Bonta said. "We take this job seriously."

The pushback is notably bipartisan. Governors and legislators in both red and blue states have resisted ceding authority, and the constitutional questions cut against the administration in places. Because Congress has not passed an express preemption statute, the White House is leaning on Article II powers and spending conditions rather than clear statutory text — a posture critics say invites Spending Clause and "unconstitutional conditions" challenges, especially where grants states already rely on are conditioned after the fact.

Why It Matters

The fight is really about who governs AI in America. If the administration prevails, a single federal standard — potentially lighter-touch than laws in California, Colorado, or Texas — would set the ceiling nationwide, and the White House's newly finalized voluntary safety-testing framework, unveiled to OpenAI, Anthropic, and Google on August 3, would become the country's dominant model of oversight. If the states prevail, developers face the very patchwork the order aims to erase, but with democratic accountability closer to the ground. At its core, the case asks whether the executive branch can achieve through litigation and funding leverage what Congress has so far declined to legislate.

What to Watch

Watch the Colorado docket, where DOJ's intervention makes xAI v. Weiser the marquee preemption test. Watch whether any state formally sues over BEAD or grant conditions, triggering the Spending Clause fight scholars anticipate. And watch Congress: a durable national framework ultimately runs through Capitol Hill, not the courts.