Last reviewed
Jun 9, 2026
Sources
20 citations
Review status
Source-backed
Revision
v2 ยท 2,692 words
Add missing citations, update stale details, or suggest a clearer explanation.
Federal preemption of state AI laws refers to the effort by the United States federal government, mainly under President Donald Trump in 2025 and 2026, to override or displace the growing body of state-level artificial intelligence statutes with a single, lighter-touch national approach. The campaign moved through several distinct phases: a proposed ten-year moratorium on state AI laws that Congress twice declined to enact, an executive order signed on December 11, 2025 directing federal agencies to challenge "onerous" state laws, the creation of a Department of Justice AI Litigation Task Force in January 2026, the first federal court intervention against a state AI law (Colorado) in April 2026, and a non-binding "National Policy Framework for Artificial Intelligence" released in March 2026 asking Congress to legislate broad preemption. Because the United States Constitution gives Congress, not the president, the authority to preempt state law, the executive-branch route has drawn sustained constitutional objections and a bipartisan backlash from state attorneys general and governors. As of mid-2026 no statute preempting state AI laws had been enacted.
In the absence of comprehensive federal AI regulation, individual states became the primary engines of AI governance in the United States. Industry groups and the Trump administration describe the result as a "patchwork" of inconsistent rules that raises compliance costs and threatens American competitiveness. Reporting in 2025 and 2026 noted that all 50 states introduced at least one AI-related bill, and trackers counted well over 1,000 AI bills nationwide.
A handful of states enacted substantive, binding obligations. The Colorado AI Act (SB 24-205), signed in May 2024, was widely described as the first comprehensive, risk-based AI accountability law in the country; it required developers and deployers of "high-risk" AI systems to use reasonable care to protect consumers from algorithmic discrimination. Its effective date was repeatedly pushed back, eventually to June 30, 2026. California enacted the Transparency in Frontier Artificial Intelligence Act (California SB 53) in September 2025, imposing risk-management disclosure, safety-incident reporting, and whistleblower protections on the largest frontier developers (those with annual revenue above $500 million); it took effect January 1, 2026. This followed California's high-profile SB 1047, a more sweeping frontier-safety bill that Governor Gavin Newsom had vetoed in 2024. Texas adopted the Responsible Artificial Intelligence Governance Act (Texas TRAIGA, HB 149), effective January 1, 2026, which barred certain intentional misuses and created a regulatory sandbox. New York advanced the Responsible AI Safety and Education Act (New York RAISE Act), focused on frontier-model safety disclosures and incident reporting. Utah and other states added narrower disclosure and consumer-protection rules. It was this body of differing, mostly recent state law that the federal preemption effort set out to displace.
The first federal vehicle was legislative. The version of the 2025 budget reconciliation package passed by the U.S. House of Representatives on May 22, 2025, commonly called the One Big Beautiful Bill Act, contained a provision that would have imposed a ten-year moratorium barring states from enforcing any law or regulation affecting AI models, AI systems, or automated decision systems.
The moratorium collapsed in the Senate. Senators Marsha Blackburn (R-Tennessee) and Maria Cantwell (D-Washington) introduced an amendment to strike the provision, and in an overnight session the Senate voted 99 to 1 to remove it on July 1, 2025. The lone vote against removal came from Senator Thom Tillis (R-North Carolina). President Trump signed the reconciliation bill into law on July 4, 2025 with no restriction on state AI legislation. The proposal had drawn opposition from across the spectrum, including a bipartisan coalition of 36 state attorneys general who wrote to congressional leaders on November 25, 2025 (after an earlier letter from a 40-attorney-general group), and a majority of Republican governors led by Arkansas Governor Sarah Huckabee Sanders.
Congress declined a second time later in 2025. A proposed AI moratorium was floated for the Fiscal Year 2026 National Defense Authorization Act, but House and Senate negotiators released a compromise NDAA without any language blocking state AI laws. Having failed in Congress twice, the administration turned to executive action.
On December 11, 2025, President Trump signed an executive order titled "Ensuring a National Policy Framework for Artificial Intelligence." Rather than preempting state law directly, which the president cannot do unilaterally, the order directed a coordinated, multi-agency campaign to undermine state AI laws through litigation, agency reinterpretation, conditional federal funding, and a request for future legislation. White House AI and crypto adviser David Sacks framed the goal narrowly, saying the administration would "push back on the most onerous examples of state regulations" while leaving child-safety rules in place.
The order assigned specific tasks and deadlines:
The order identified three legal theories for challenging state laws: that they unconstitutionally regulate interstate commerce, that they are preempted by existing federal regulation, or that they are "otherwise unlawful" in the Attorney General's judgment.
The executive order's litigation arm took shape quickly. On January 9, 2026, Attorney General Pam Bondi issued a memorandum to the Department of Justice formally establishing the AI Litigation Task Force, to be led by Bondi or her appointee and staffed from the Civil Division, the Solicitor General's office, and the offices of the Deputy and Associate Attorney General, in consultation with Sacks. The task force was charged with challenging state AI laws on the three grounds named in the order, including the dormant Commerce Clause and federal preemption.
The first concrete action followed in April 2026 and centered on Colorado. On April 9, 2026, Elon Musk's xAI filed suit in the U.S. District Court for the District of Colorado to enjoin the Colorado AI Act, arguing it compelled speech and discriminated by content and viewpoint in violation of the First Amendment, impermissibly burdened interstate commerce, was unconstitutionally vague, and violated equal protection. On April 24, 2026, the DOJ moved to intervene in support of xAI, the first time the federal government sought to invalidate a state AI law. The same day, xAI and the Colorado Attorney General jointly moved to suspend enforcement pending the legislature's 2026 session and a ruling on a preliminary injunction; the court granted that motion on April 27, 2026, temporarily halting enforcement. Colorado's legislature then passed SB 26-189, repealing and replacing the AI Act with a narrower regime governing automated decision-making technology, and Governor Jared Polis signed it on May 14, 2026.
Separately, the administration delivered the legislative proposal contemplated by the order. On March 20, 2026, the White House released its National Policy Framework for Artificial Intelligence, a set of recommendations urging Congress to enact broad federal preemption of state AI laws deemed unduly burdensome or inconsistent with federal policy, and to bar states from regulating AI model development or imposing liability on developers for third parties' unlawful use of their systems. The framework was explicitly not binding; commentators described it as a legislative blueprint rather than enacted law. It preserved several categories of state authority, including generally applicable laws (fraud, consumer protection, and child safety), state procurement and zoning powers, and a state's governance of its own use of AI. Notably, it did not call for a new federal AI regulator, favoring oversight through existing agencies, sector-specific rules, and industry-led standards.
The most contested tool in the December order was financial: conditioning federal broadband money on a state's AI laws. The Broadband Equity, Access, and Deployment (BEAD) program, created in the 2021 infrastructure law, totals roughly $42.45 billion. The order targeted "non-deployment" funds (money for activities such as digital-literacy and workforce programs rather than physical network buildout), which a 2025 administration restructuring had left as roughly $21 billion outside the core state allocations. Section 5 instructed the National Telecommunications and Information Administration to issue a notice making states with "onerous" AI laws ineligible for those funds.
The legal vulnerability of this approach drew immediate attention. As analysts at Lawfare noted, the BEAD statute is about deploying broadband and connecting locations and never mentions AI, which exposes the strategy to the major-questions doctrine and to federalism clear-statement rules requiring that grant conditions and any displacement of traditional state authority be unambiguous. Under the Spending Clause framework of South Dakota v. Dole, conditions must relate to the federal interest in the funded program, a "relatedness" hurdle the AI condition may not clear. Coercion was viewed as a weaker objection, since the sums at stake represent only about 2 to 3 percent of most state budgets, below the threshold the Supreme Court flagged in NFIB v. Sebelius. The leverage also met bipartisan political resistance: Commerce Secretary Howard Lutnick told Congress in February 2026 that states would keep their full BEAD allocations "in accordance with federal law," saying "We will hew to the law for sure," while conceding he was "not sufficiently familiar" with how the order squared with BEAD law. Democratic governors, including Illinois Governor JB Pritzker, publicly pressed for the release of frozen broadband funds.
The preemption campaign split along industry-versus-states lines rather than cleanly by party. Major AI firms and some venture investors backed a single national standard, arguing the patchwork is unworkable for companies operating nationwide. Opposition was unusually bipartisan. The 99-to-1 Senate vote, the 36-attorney-general letter, and the GOP-governor letter signaled that a sweeping federal override was politically difficult, and several Republican governors, including Sanders, Ron DeSantis of Florida, and Spencer Cox of Utah, objected on states-rights grounds. DeSantis publicly observed that Congress could preempt states through legislation but suggested an executive order alone could not. State attorneys general also pushed back at the agency level, with a bipartisan group of 23 filing comments on December 17, 2025 opposing FCC preemption of state AI laws.
The central legal question is structural: only Congress can preempt state law, so an executive order cannot itself nullify a state statute. The order's force therefore depends on winning lawsuits, persuading agencies to act, and surviving challenges to the funding conditions. Each theory faces obstacles. The dormant Commerce Clause argument is complicated by National Pork Producers Council v. Ross (2023), in which the Supreme Court declined to strike a nondiscriminatory state law merely because it forced out-of-state businesses to change practices. The First Amendment theory, aimed at laws touching AI "outputs," is novel and untested. And the BEAD funding strategy confronts the statutory and Spending Clause problems described above. Law-firm analyses broadly predicted the order would be litigated and that its practical effect on most state laws would be limited unless Congress acts. The net result through mid-2026 was an unresolved standoff: state laws remained on the books (Colorado's a partial exception, narrowed by its own legislature), the federal government had opened a single test case, and the durable question of nationwide preemption sat with Congress, which had so far declined to answer it.
| Date | Action | Status |
|---|---|---|
| May 2024 | Colorado enacts the Colorado AI Act (SB 24-205) | Enacted (effective date later delayed to June 30, 2026) |
| May 22, 2025 | House passes reconciliation bill with a 10-year state AI moratorium | Proposal |
| July 1, 2025 | Senate strikes the moratorium 99-1 (Blackburn-Cantwell amendment) | Proposal defeated |
| July 4, 2025 | Trump signs the One Big Beautiful Bill Act with no AI moratorium | Enacted (without moratorium) |
| September 2025 | California enacts SB 53 (frontier AI transparency) | Enacted (effective Jan 1, 2026) |
| Late 2025 | AI moratorium dropped from the FY2026 NDAA | Proposal defeated |
| December 11, 2025 | Trump signs EO "Ensuring a National Policy Framework for Artificial Intelligence" | Executive order |
| December 17, 2025 | 23 state attorneys general file comments opposing FCC preemption | Opposition |
| January 9, 2026 | AG Bondi establishes the DOJ AI Litigation Task Force | Agency action |
| February 2026 | Commerce Secretary Lutnick tells Congress states keep full BEAD funds | Agency statement |
| March 20, 2026 | White House releases the National Policy Framework for AI | Non-binding recommendation |
| April 9, 2026 | xAI sues to enjoin the Colorado AI Act | Litigation filed |
| April 24, 2026 | DOJ moves to intervene against the Colorado AI Act | Litigation (first federal challenge) |
| April 27, 2026 | Court suspends enforcement of the Colorado AI Act | Court order (temporary) |
| May 14, 2026 | Governor Polis signs SB 26-189 narrowing/replacing the Colorado AI Act | Enacted |
Related: AI regulation, AI executive order, AI Action Plan, export controls.