Last reviewed
May 31, 2026
Sources
32 citations
Review status
Source-backed
Revision
v3 ยท 3,222 words
Add missing citations, update stale details, or suggest a clearer explanation.
The federal AI preemption executive order is Executive Order 14365, "Ensuring a National Policy Framework for Artificial Intelligence," which President Donald Trump signed on December 11, 2025. The order sets out a national policy of a "minimally burdensome" approach to artificial intelligence and directs federal agencies to push back against state AI laws the administration considers onerous. Its main tools are a Department of Justice litigation task force charged with challenging state AI laws in court, an evaluation by the Department of Commerce to flag conflicting state laws, conditions on certain federal funding such as the broadband BEAD program, and a request for federal legislation that would preempt state rules. The Federal Register published the order on December 16, 2025, at 90 FR 58499.[1][2]
The order followed a failed attempt earlier in 2025 to put a moratorium on state AI laws into federal budget legislation, and a second failed attempt in late 2025 to attach preemption language to the annual defense bill. It does not by itself repeal any state law. As several law firms noted after the signing, only Congress or the courts can actually invalidate a state statute, so the order works through litigation, agency action, and funding pressure rather than direct preemption.[3][4] This article describes what the order does, dates each piece precisely, and lays out the federalism debate and the reactions it drew.
The administration's direction on AI was set out in America's AI Action Plan, released on July 23, 2025. The plan treated rapid AI development as a national priority and called for removing regulatory barriers. One recommendation asked the Office of Management and Budget and federal agencies to weigh a state's AI regulatory climate when making certain discretionary funding decisions, so that federal money would not flow to states whose rules the administration saw as hindering that funding. Critics read this as an early move to discourage state regulation through federal leverage, and the December order later put a version of the idea into operation.[5][2] For broader context on presidential AI directives, see Executive Order on AI.
Earlier in 2025, Congress took a more direct route. The version of the budget reconciliation bill that passed the House of Representatives, the measure known as the One Big Beautiful Bill Act, contained a provision barring states from enforcing their own AI laws for ten years. The provision drew opposition from a broad coalition of state officials, consumer groups, labor unions, and faith organizations. Senators Marsha Blackburn and Ted Cruz worked on a compromise that would have shortened the timeline and tied it to broadband funding, but Blackburn backed out within a day of striking the deal. On July 1, 2025, during an overnight session, the Senate voted 99 to 1 to strip the moratorium from the bill, with Senator Thom Tillis casting the only vote against the amendment.[6][7][8] The amendment was offered by a bipartisan group including Blackburn and Senate Commerce ranking member Maria Cantwell. The National Conference of State Legislatures, which had lobbied against the measure, welcomed the result.[9]
With the budget route closed, supporters of preemption tried again in late 2025 by seeking to add language to the National Defense Authorization Act for fiscal year 2026. That effort also failed. Bipartisan coalitions of state attorneys general and state legislators opposed it, and the chairs and ranking members of the armed services committees did not include it.[10][3] The executive order came after that second defeat, as the administration shifted to tools it could use without new legislation.
The order declares a national policy, in its words, to "sustain and enhance the United States' global AI dominance through a minimally burdensome national policy framework for AI," and frames a growing patchwork of state laws as a threat to that goal. The White House fact sheet said state legislatures had introduced more than 1,000 AI bills. The substantive sections then assign tasks to specific agencies with deadlines.[2][11]
The most discussed provision creates a Department of Justice AI Litigation Task Force. The order directs the Attorney General to establish the task force within 30 days, with the job of challenging state AI laws the administration views as inconsistent with the order's policy, on grounds such as unconstitutional regulation of interstate commerce, preemption by existing federal law, or that the laws are otherwise unlawful. The Attorney General formally established the task force by memorandum on January 9, 2026.[12][1] Legal analysts noted that at least one likely theory rests on the dormant Commerce Clause, the doctrine that limits states from unduly burdening interstate commerce.[13]
The order directs the Secretary of Commerce to publish, within 90 days, an evaluation identifying state AI laws that conflict with the national policy, including laws that require AI models to alter their truthful outputs and laws with disclosure or reporting requirements the administration considers problematic. The evaluation is meant to flag laws for the task force or for other action. Commerce Secretary Howard Lutnick was named to lead that review.[1][14] The reference to laws that "require alterations to the truthful outputs of AI models" connects to the administration's earlier interest in what it called ideologically biased or "woke" AI.[14]
Funding leverage runs through the Department of Commerce as well. The order directs the Assistant Secretary for Communications and Information, who heads the National Telecommunications and Information Administration, to issue a policy notice within 90 days specifying that states with onerous AI laws are ineligible for certain non-deployment funds under the Broadband Equity, Access, and Deployment program, to the maximum extent federal law allows. Analysts estimated that roughly 21 billion dollars in BEAD non-deployment funds could be affected.[15][16] The choice of BEAD echoed the earlier Blackburn and Cruz compromise idea of linking broadband money to state AI rules.
Two more agencies received tasks. The order directs the Chairman of the Federal Communications Commission to consider initiating a proceeding on a federal reporting and disclosure standard for AI models that could preempt conflicting state requirements. It directs the Chairman of the Federal Trade Commission to issue a policy statement, within 90 days, on how Section 5 of the FTC Act, which bars unfair and deceptive practices, applies to AI, and when state laws that require alterations to truthful AI outputs would be preempted.[1][17] Finally, the order asks the Special Advisor for AI and Crypto, David Sacks, together with the Assistant to the President for Science and Technology, Michael Kratsios, to prepare a legislative recommendation for a uniform federal framework that would preempt conflicting state laws.[2][14]
The order carves out several categories of state law from its preemption push. It states that it does not target laws addressing child safety, AI compute and data center infrastructure (apart from generally applicable permitting reforms), and state government procurement and use of AI, along with other topics to be determined later. The signed order singled out Colorado's algorithmic discrimination law as an example, arguing it could push AI models to produce false results in order to avoid differential treatment of protected groups. An earlier draft that leaked in November 2025 had named California's frontier AI transparency law, but that reference was dropped from the final text.[11][18]
The table below summarizes the directives, the responsible agency, and the deadlines stated in the order. It reflects the order as signed, not later implementation, which lagged in several cases.
| Directive | Agency or actor | Deadline (from EO) |
|---|---|---|
| Establish AI Litigation Task Force to challenge state AI laws | Attorney General / DOJ | Within 30 days (established Jan 9, 2026) |
| Publish evaluation of conflicting state AI laws | Secretary of Commerce (Howard Lutnick) | Within 90 days |
| Policy notice on BEAD non-deployment funds for states with onerous AI laws | NTIA (Assistant Secretary for Communications and Information) | Within 90 days |
| Consider proceeding on federal AI reporting and disclosure standard | FCC Chairman | Discretionary |
| Issue policy statement on FTC Act and AI output laws | FTC Chairman | Within 90 days |
| Prepare legislative recommendation for a uniform federal framework | David Sacks and Michael Kratsios | Not fixed (published Mar 20, 2026) |
The order is signed and in effect, but agencies moved at different speeds, and several deadlines passed without the underlying actions being taken. The Justice Department established the litigation task force by memorandum on January 9, 2026, though reporting in the following months indicated it had not yet filed suit against any state AI law.[12][19] The Commerce Department evaluation of state laws was expected in spring 2026 rather than within the literal 90-day window, and the FTC policy statement had not been issued as of early April 2026 according to the International Association of Privacy Professionals.[19] On March 20, 2026, the White House published the legislative recommendation as the National Policy Framework for Artificial Intelligence, which urged Congress to adopt a light-touch federal regime and to preempt conflicting state laws.[20][21]
Because the order cannot itself strike down a state statute, law firms advised companies to keep complying with applicable state AI laws while the legal questions play out.[4][22] State laws remained enforceable, and the practical effect of the order depended on whether litigation succeeded and whether Congress acted.
The central legal question is whether the executive branch can displace state AI laws on its own. Preemption usually flows from federal statute, where Congress either says explicitly that federal law overrides state law or builds a scheme comprehensive enough that courts infer it. An executive order cannot by itself repeal a state statute. What it can do, as analyses by the Brennan Center and others described, is direct agencies and the Justice Department to argue in court that particular state laws are preempted by existing federal authority, and to use funding and regulatory tools the executive branch already controls.[23][13] Several commentators were skeptical that broad preemption claims would succeed without a clear federal statute. A Lawfare analysis argued that the BEAD funding strategy in particular faced steep statutory hurdles, since the BEAD law does not plainly let the administration condition the money on a state's AI policy, which gave state challengers what the author called the better reading of the statute.[13]
The patchwork concern is the main argument for a national framework. Industry groups and the administration say that dozens of differing state laws on transparency, automated decision systems, and deepfakes raise compliance costs that fall hardest on smaller developers and slow national deployment. The order and fact sheet leaned on the count of more than 1,000 state AI bills as evidence of fragmentation.[2][11] Opponents cite the same volume as evidence that states are responding to real problems their residents face, and that removing state laws without enacting federal protections in their place would leave a gap rather than a clean single standard.[24][25]
Reaction split along lines that did not track the usual partisan divide. The clearest opposition from states came from officials who said they would defend their laws in court. Colorado Attorney General Phil Weiser indicated the state would challenge the order, and California state senator Scott Wiener, the author of the state's frontier AI law, called it a handout to large AI companies and said the state would fight it.[25][26] Several Republican officials also pushed back on states' rights grounds. Florida Governor Ron DeSantis said his state had the right to regulate AI, putting it plainly: "We have a right to do this." Arkansas Governor Sarah Huckabee Sanders, who had earlier led a group of 17 Republican governors against the budget moratorium, again urged the administration to drop the preemption plan.[27][28] Representative Marjorie Taylor Greene publicly criticized the approach as a threat to state authority, and reporting described the order as dividing the president's own party.[28]
Industry voices were more favorable. David Sacks, the administration's AI advisor, framed the effort as targeting only the most burdensome state rules, saying the administration would protect child safety but would push back on the most onerous examples of state regulation.[14] Many large technology companies had supported a single national standard over a patchwork.
Civil society groups largely opposed the order. The Center for Democracy and Technology said the order did nothing to address the documented harms AI systems can cause, and a coalition that included Consumer Reports, the Electronic Privacy Information Center, and the Lawyers' Committee for Civil Rights Under Law argued that preempting state laws while no comprehensive federal law exists would strip consumer and civil rights protections.[25][29] During the earlier NDAA fight, bipartisan coalitions of 36 state attorneys general and 290 state legislators had opposed adding preemption language to the defense bill, and many of the same groups carried that opposition into the debate over the order.[10][29]
A specific focus of the broader preemption push, in the view of several analysts, was California. The state enacted California Senate Bill 53, the Transparency in Frontier Artificial Intelligence Act, which Governor Gavin Newsom signed on September 29, 2025. SB 53 requires large frontier AI developers to publish safety frameworks and to report certain serious incidents, with most requirements taking effect January 1, 2026. The leaked November draft of the order had singled out SB 53, and although the final order dropped that reference in favor of Colorado, commentators still described SB 53 as the kind of state law the litigation task force could test, since it imposes obligations on companies that operate nationwide.[18][30] The Colorado Artificial Intelligence Act and the Texas Responsible AI Governance Act figured in the same debate as further examples of the patchwork the administration cited.
The contrast with Europe sharpened the argument on both sides. The EU AI Act, the European Union's risk-based AI law, is a centralized regulatory approach, and supporters of a single United States framework sometimes pointed to the value of one standard rather than fifty. Critics noted that Europe's approach adds regulation rather than removing it, so the comparison cut both ways depending on whether the goal was uniformity or deregulation. By late 2025 the EU was itself weighing changes to the timing of parts of the AI Act through its Digital Omnibus proposal, published on November 19, 2025, which fed into United States arguments about how prescriptive any framework should be.[31][32] The dispute connects to the wider fields of AI regulation and AI governance, where the balance between national rules and subnational experimentation is a recurring question.
The order matters as a strategy as much as for any single provision. After preemption failed twice in Congress, in the budget bill and then in the defense bill, the administration turned to executive tools it could deploy alone: litigation, agency review, and funding conditions. The early pushback, including from Republican governors and a Republican member of Congress, showed that preemption of state AI law is contested across party lines, with some conservatives defending state authority and some industry voices favoring a national rule.[10][28]
The episode also marked out the limits of executive action here. Because durable preemption generally needs a statute, analysts concluded that the order would likely run alongside, rather than replace, a renewed push for a federal AI law, which is why the March 2026 legislative framework matters. Heading further into 2026, policy trackers described the preemption fight as unresolved, with states continuing to legislate and to enforce their laws, the litigation task force not yet having sued, and Congress still weighing whether to pass a national standard.[19][21] How that resolves will shape whether AI in the United States is governed mainly from Washington, mainly from state capitals, or through some negotiated mix of the two.