What the order actually directs
The order came from the White House on December 11, 2025, and it directs the Attorney General to stand up an AI Litigation Task Force inside the Department of Justice within 30 days, putting the task force's mandate in effect starting January 10, 2026. Its sole job is to challenge state AI laws in federal court on one of three grounds: that they unconstitutionally burden interstate commerce, that they are preempted by existing federal regulation, or that they are otherwise unlawful in the Attorney General's own judgment. [1]
Beyond litigation, the order leans on federal money and rulemaking. The Commerce Secretary has 90 days to publish an evaluation naming state AI laws the administration considers "onerous," and states that land on that list become ineligible for non-deployment funds under the Broadband Equity Access and Deployment program "to the maximum extent allowed by Federal law." Other agencies are told to examine whether they can condition their own discretionary grants the same way. Separately, the order directs the FCC to consider a federal AI reporting and disclosure standard meant to preempt conflicting state rules, and the FTC to issue a policy statement within 90 days on how its existing unfair-and-deceptive-practices authority applies to AI, while senior administration officials draft legislative recommendations for Congress to enact a uniform federal framework. [1][3]
Why the order could not simply preempt state law
Atlas interpretation: An executive order is not a statute, and the Supremacy Clause lets federal law override state law only when Congress has passed a statute or delegated regulatory authority that reaches the conflict. Neither existed here: there is no general federal AI law for the order to derive preemptive force from, and Congress had already declined to write one. In July 2025 the Senate voted 99 to 1 to strip a ten-year moratorium on state AI regulation out of that year's budget reconciliation bill. Even OpenAI's own March 2025 submission asking the White House Office of Science and Technology Policy for preemption relief had conceded the point in a footnote: "Federal preemption over existing or prospective state laws will require an act of Congress." The order works around that limit by having agencies sue, condition money, and write competing rules, rather than purporting to preempt anything by presidential signature alone. [3][7]
Atlas interpretation: Legal analysts were skeptical the workaround would hold up. Gibson Dunn concluded that "the practical impact of the EO in the short- and medium-term is likely to be limited," since most of the text instructs agencies to act rather than changing anyone's legal obligations on its own, leaving courts little to rule on until an agency actually files suit or denies a grant. The interstate commerce theory faces its own trouble: state AI laws generally regulate conduct occurring within the state rather than discriminating against out-of-state business, and the Supreme Court's 2023 National Pork Producers Council decision narrowed how far that doctrine reaches. The funding conditions face a separate test, from South Dakota v. Dole, requiring that conditions be germane to the program and not so large a share of funding as to be coercive; case law on where that line sits is thin, and in analogous fights over sanctuary-city funding it produced split appellate rulings. As one Carnegie Endowment analysis summarized it, the order "gives agencies a list of tasks. It doesn't really do anything by itself." [2][3][4]
SB 53, Colorado, and the states that pushed back
The signed order names one law outright: Colorado's algorithmic discrimination statute, which it argues risks forcing AI systems to alter "truthful outputs" in violation of the First Amendment. California's law is not named as explicitly in the final text as it reportedly was in an earlier draft that circulated in November 2025. Legal analysts nonetheless read SB 53, with its mandatory safety-framework publication and incident reporting, and New York's pending RAISE Act as the clear implied targets, since both impose exactly the kind of frontier-model disclosure duties the order's "minimally burdensome" framing argues against. [1][4]
State government's reaction was fast and bipartisan in its anger if not its politics. California's Senator Alex Padilla said the administration was "attacking state leadership and basic safeguards in one fell swoop." The opposition also drew on an existing coalition: more than 260 state legislators from all 50 states had already called an earlier federal preemption push "reckless" in a June 2025 letter, months before this order was signed. Civil society groups were similarly blunt. The ACLU's Cody Venzke called the funding threats "dangerous" and likely unconstitutional, the Future of Life Institute's Michael Kleinman described the order as "a gift for Silicon Valley oligarchs," and the American Federation of Teachers' Randi Weingarten called it "outrageous and a likely illegal directive." Officials in both California and Colorado signaled they would challenge the order in court. [5]
Industry lobbying, and OpenAI's own use of SB 53
Industry reaction split along predictable lines. The U.S. Chamber of Commerce's Michael Richards argued that a patchwork of state AI rules "will stifle profits and undermine the nation's innovation and leadership," while the Security Industry Association's Jake Parker welcomed the order as aimed at broad regulatory regimes rather than narrowly scoped state statutes. That patchwork argument was not new. It is the same one Meta, OpenAI, and investors including Andreessen Horowitz made against SB 53 while it moved through the California legislature in 2025, when they urged Governor Newsom to veto it and wait for federal rules instead. [6]
Atlas interpretation: OpenAI's own subsequent record complicates that argument. Having lobbied against SB 53 as an example of unwanted state regulation, and having told federal officials as early as March 2025 that only Congress could actually preempt state law, the company reversed course on the California bill itself: by August 2026 it was publicly asking California to strengthen SB 53's requirements rather than repeal them. Around the same time, OpenAI proposed that Congress build a national frontier-AI framework modeled explicitly on SB 53, New York's RAISE Act, and Illinois's SB 315, then use that federal framework to preempt whatever state rules diverged from it, an approach one report labeled "reverse federalism." The company that spent 2025 arguing a patchwork of state AI laws should yield to a single federal rule ended up treating one state's law as the federal rule's rough draft. [7][8]
Sources
- Ensuring a National Policy Framework for Artificial Intelligence
The White House · Dec 11, 2025
- President Trump's Latest Executive Order on AI Seeks to Preempt State Laws
Gibson Dunn · Dec 15, 2025
- Executive Order Takes Aim at State AI Laws
Morrison Foerster · Dec 13, 2025
- The Most Likely Outcomes of Trump's Order Targeting State AI Laws
Carnegie Endowment for International Peace · Dec 15, 2025
- State leaders, civil rights groups respond to 'dangerous' Trump order banning state AI laws
StateScoop · Dec 12, 2025
- Trump's state AI-law order sparks clash between states and industry
StateScoop · Dec 18, 2025
- OpenAI Wants Congress to Preempt State AI Safety Laws
Implicator.ai · Jun 3, 2026
- OpenAI says California should strengthen its AI safety bill
TechCrunch · Aug 22, 2026