TL;DR: On December 11, 2025, President Trump signed Executive Order 14365 targeting state AI laws. The order cannot directly preempt state legislation — only Congress can do that — but it deploys $20B in funding leverage, a DOJ AI Litigation Task Force, and binding grant conditions to pressure states into compliance. California, Colorado, and New York are the primary targets. Both Congressional attempts at statutory preemption have already failed. The result is a high-stakes jurisdictional standoff with no clear legal resolution in sight.
The battle over who gets to regulate artificial intelligence in the United States is no longer hypothetical. Since December 2025, the Trump administration has been systematically pushing back against state-level AI oversight, deploying executive authority, funding leverage, and litigation threats to assert federal supremacy over a domain that dozens of states were rapidly claiming as their own.
The question at stake is not technical. It is constitutional. Who has the right to set the rules for artificial intelligence in America — Washington, or the fifty states? The answer, it turns out, is genuinely contested. And the current administration is determined to settle it in favor of the federal government, by any means available short of legislation.
The Executive Order and What It Actually Does
The centerpiece is Executive Order 14365, signed December 11, 2025, officially titled "Ensuring a National Policy Framework for Artificial Intelligence." The order does not itself preempt state AI laws — only an act of Congress can do that. But it deploys three indirect pressure mechanisms designed to achieve the same result in practice.
Keep Your AI Law, Lose the Grant
The Commerce Department was directed to publish a list of "onerous" state AI laws within 90 days. States on that list become ineligible for roughly $20 billion in non-deployment BEAD broadband program funds. For state governments that had been counting on those infrastructure dollars, the financial pressure is immediate and significant.
A Contractual Suspension of State Authority
States seeking any form of discretionary federal funding must sign binding agreements not to enforce their AI laws during the entire grant performance period. This is not a temporary waiver or a policy preference — it is a contractual suspension of state legislative authority delivered through the backdoor of grant conditions.
Taking States to Court
The Attorney General was directed to establish a DOJ AI Litigation Task Force within 30 days, tasked with challenging state AI laws on three constitutional grounds: the Dormant Commerce Clause (arguing state laws burden interstate commerce), FTC Act preemption, and First Amendment conflicts for laws mandating AI disclosures or output alterations.
Which States and Laws Are Targeted
Colorado's AI Act is the only state law singled out by name in the EO. The order frames it as forcing AI models to "embed ideological bias" by requiring companies to audit algorithms for discriminatory impact on protected groups. That framing is the administration's primary justification for the entire preemption effort.
Beyond Colorado, the following laws are directly in the crosshairs:
- California SB 53 — the Transparency in Frontier AI Act, requiring large developers to publish annual frontier AI governance frameworks and catastrophic risk documentation.
- California AB 2013 — the Generative AI Training Data Transparency Act, requiring detailed disclosure of training data sources. Effective January 1, 2026.
- New York RAISE Act — requires frontier model developers to publish safety protocols and conduct pre-deployment risk assessments.
In total, more than 40 states had introduced some form of AI legislation heading into 2026. The EO's Commerce Department list — once published — is expected to encompass a broad range of these efforts, not just the three states named above.
Why the Legal Strategy Is Contested
The administration's three constitutional theories each face significant obstacles in court. Legal analysts at Gibson Dunn and Latham & Watkins have flagged the Dormant Commerce Clause argument as substantially weakened by the Supreme Court's 2023 ruling in National Pork Producers v. Ross, which gave states considerably more latitude to regulate commerce within their borders than the administration's theory assumes.
The FTC Act preemption theory faces a different problem: an FTC policy statement lacks the force of law required to override duly enacted state statutes. Federal preemption through agency guidance, rather than statute, is a legally fragile position.
The First Amendment theory — arguing that disclosure mandates violate free speech protections — is the most novel of the three, and also the least tested in the context of AI regulation specifically.
Congress Failed Twice — and That Changes Everything
What gives the legal fragility of EO 14365 real weight is the context in which it operates. The administration did not turn to executive pressure as a first resort. Congressional attempts at hard statutory preemption — including a proposed 10-year moratorium on state AI laws inserted into a reconciliation bill, and a separate provision in the National Defense Authorization Act — both failed in 2025.
Those failures matter enormously. Without a statute, there is no clean legal mechanism through which federal AI policy can override state law. The EO is working around a democratic failure, not alongside a democratic success. That distinction will matter in court.
A follow-on National AI Policy Framework, released March 20, 2026, reiterated the call for Congress to enact uniform federal standards — but carries no legal force of its own. It is, in effect, a wish list.
TechVernia Verdict
Executive Order 14365 is the most aggressive assertion of federal AI authority in US history — and also, potentially, one of the most legally exposed. The administration has correctly identified a real problem: a patchwork of fifty different state AI regimes would create genuine compliance chaos for developers operating at national scale. But the solution it has chosen bypasses the one mechanism that could actually deliver legal clarity: legislation.
The result is a standoff. State AI laws remain enforceable. The DOJ task force is building its cases. States are watching their federal funding pipelines with anxiety. The jurisdictional battle is not just beginning — it may define the governance of AI in America for the next decade.
Frequently Asked Questions
No — not directly. Under the US constitutional system, only an act of Congress can preempt state law through the Supremacy Clause. An executive order can direct federal agencies to challenge state laws in court and to condition federal funding on non-enforcement, but it cannot itself nullify legislation passed by a state legislature. This is the central legal vulnerability of EO 14365.
Colorado is the only state named in the EO text. California and New York are the most prominently targeted through the broader enforcement mechanisms, given the scope of their AI legislation (SB 53, AB 2013, and the RAISE Act respectively). However, the Commerce Department's "onerous laws" list — once published — is expected to cover AI legislation from dozens of states.
The BEAD (Broadband Equity, Access, and Deployment) program is a $42.45 billion federal broadband initiative. EO 14365 makes states with AI laws on the Commerce Department's "onerous" list ineligible for approximately $20 billion of non-deployment BEAD funds. For states that had planned their broadband infrastructure rollouts around this money, the financial pressure is substantial and immediate.
Several things are likely in parallel: the DOJ task force will file its first legal challenges, states will mount Spending Clause defenses against the funding conditions, and the Commerce Department will publish its "onerous laws" list — which will trigger its own round of litigation. Meanwhile, Congress may attempt a third round of statutory preemption legislation in 2026. The legal battles could take years to resolve, leaving AI developers in an extended period of regulatory uncertainty.
Related Articles:
