Executive Order 14365

Executive Order 14365, "Ensuring a National Policy Framework for Artificial Intelligence," is an executive order signed by President Donald Trump on December 11, 2025 and published in the Federal Register on December 16, 2025 at 90 FR 58499 (Executive Order 14365 — Ensuring a National Policy Framework for AI). It declares a policy of a "minimally burdensome national policy framework for AI" and directs the federal government to challenge and displace state AI laws through a Department of Justice AI Litigation Task Force, a Commerce Department evaluation of state laws, conditions on federal broadband and discretionary grant funding, Federal Communications Commission and Federal Trade Commission actions, and a legislative recommendation for a preemptive federal framework. The order seeks to prevent state regulation rather than to create new federal regulatory obligations, and it is the executive branch's principal instrument in the federal–state dispute over AI regulation.

Status and timeline

The order set dated deliverables that fell due in the first quarter of 2026. The table below lists each against its statutory deadline and what has been reported since.

Stated purpose

Section 1 frames the order as a response to state AI regulation that, in the administration's account, threatens United States leadership in AI. It gives three grounds: state-by-state regulation "creates a patchwork of 50 different regulatory regimes" that burdens compliance, particularly for start-ups; state laws are "increasingly responsible for requiring entities to embed ideological bias within models"; and state laws "sometimes impermissibly regulate beyond State borders." As its example of the second ground, it cites "a new Colorado law banning 'algorithmic discrimination'" (the Colorado AI Act), which it says "may even force AI models to produce false results in order to avoid a 'differential treatment or impact' on protected groups" (Executive Order 14365 — Ensuring a National Policy Framework for AI).

The order calls for a national standard, developed with Congress, that would "forbid State laws that conflict with the policy set forth in this order" while ensuring that children are protected, censorship is prevented, copyrights are respected, and communities are safeguarded. Until such a standard exists, it directs the administration to act against "the most onerous and excessive laws emerging from the States." Section 2 states the policy as sustaining and enhancing "the United States' global AI dominance through a minimally burdensome national policy framework for AI" (Executive Order 14365 — Ensuring a National Policy Framework for AI).

Nita Farahany notes that the order defines neither "AI" nor "onerous" (Inside My AI Law & Policy Class 25 (Special Edition): The Executive Order That Could Kill State AI Laws (Farahany, December 2025)).

Key provisions

The order's operative sections establish five mechanisms: litigation, evaluation and designation, funding conditions, agency preemption actions, and a legislative recommendation (Executive Order 14365 — Ensuring a National Policy Framework for AI; Inside My AI Law & Policy Class 25 (Special Edition): The Executive Order That Could Kill State AI Laws (Farahany, December 2025)).

AI Litigation Task Force (§ 3). Within 30 days, the Attorney General must establish a task force whose "sole responsibility" is to challenge state AI laws inconsistent with the § 2 policy, including on grounds that they unconstitutionally regulate interstate commerce, are preempted by existing federal regulations, or are otherwise unlawful in the Attorney General's judgment, and including laws identified under § 4. The Task Force consults the Special Advisor for AI and Crypto, the Assistant to the President for Science and Technology, the Assistant to the President for Economic Policy, and the Counsel to the President on which state laws warrant challenge.

Evaluation of state AI laws (§ 4). Within 90 days, the Secretary of Commerce, acting under 47 U.S.C. 902(b) and in consultation with the same White House officials, must publish an evaluation identifying onerous state laws that conflict with the § 2 policy and laws that should be referred to the Task Force. At a minimum the evaluation must identify laws that "require AI models to alter their truthful outputs" and laws that may compel developers or deployers to disclose or report information in a manner that would violate the First Amendment or another constitutional provision. It may also identify state laws that promote AI innovation consistent with the policy.

Restrictions on state funding (§ 5). Within 90 days, the Secretary of Commerce, through the Assistant Secretary for Communications and Information (NTIA), must issue a Policy Notice on eligibility for remaining Broadband Equity, Access and Deployment (BEAD) funds. The notice must make states with onerous AI laws identified under § 4 ineligible for non-deployment funds "to the maximum extent allowed by Federal law," and must describe how a fragmented state regulatory landscape threatens BEAD-funded deployments. Separately, agencies must assess, in consultation with the Special Advisor for AI and Crypto, whether they may condition discretionary grants on states either not enacting conflicting AI laws or, for states that have enacted them, entering a binding agreement not to enforce those laws during the funding performance period. The funding conditions can operate before any court rules on whether a state law is preempted.

Federal reporting and disclosure standard (§ 6). Within 90 days of publication of the § 4 identification, the FCC Chairman must initiate a proceeding to determine whether to adopt a federal reporting and disclosure standard for AI models that preempts conflicting state laws.

Preemption of state laws mandating deceptive conduct (§ 7). Within 90 days, the FTC Chairman must issue a policy statement on the application of the FTC Act's prohibition on unfair and deceptive acts or practices (15 U.S.C. 45) to AI models, explaining when state laws that require alterations to the truthful outputs of AI models are preempted.

Legislative recommendation (§ 8). The Special Advisor for AI and Crypto and the Assistant to the President for Science and Technology must jointly prepare a legislative recommendation establishing a uniform federal policy framework for AI that preempts conflicting state AI laws. Under § 8(b), the recommendation "shall not propose preempting otherwise lawful State AI laws" relating to (i) child safety protections; (ii) AI compute and data center infrastructure, other than generally applicable permitting reforms; (iii) state government procurement and use of AI; and (iv) other topics to be determined.

General provisions (§ 9). The order does not impair existing agency authority or OMB functions, is to be implemented consistent with applicable law and subject to appropriations, creates no enforceable right or benefit, and assigns publication costs to the Department of Commerce.

State laws in scope

Beyond its reference to Colorado, the order names no state statute. The § 4 criterion covering laws that compel developers to disclose or report information reaches state frontier-transparency laws such as California SB 53 and the New York RAISE Act (Executive Order 14365 — Ensuring a National Policy Framework for AI). California SB 243 is among the other state AI laws exposed to the order's mechanisms.

Scope of the state-procurement carve-out

The § 8(b) carve-outs are written as limits on what the legislative recommendation may propose; the Task Force mandate in § 3 and the funding and agency provisions in §§ 5–7 contain no equivalent exclusion (Executive Order 14365 — Ensuring a National Policy Framework for AI). Farahany reads the child-safety exemption the same way, as applying only to the legislative-recommendation mechanism and not to DOJ litigation, BEAD funding loss, or agency rulemaking, so that state child-safety laws could still face Task Force challenges (Inside My AI Law & Policy Class 25 (Special Edition): The Executive Order That Could Kill State AI Laws (Farahany, December 2025)).

Andrew Clearwater reads Section 8 as carving state procurement and governmental use of AI out of the order's preemption scope, and describes California Governor Newsom's EO N-5-26 (March 30, 2026) as "finding the legal seam in the federal preemption wall" (California Just Played the One AI Card the Feds Can't Block (Clearwater, March 2026)). On that reading, the procurement carve-out stands out against the order's otherwise broad claim, and it makes EO N-5-26 a workable counter-instrument because that order operates through state purchasing rather than direct regulation of AI developers. The argument links the order to Procurement-Driven AI Governance.

Regulatory approach

The order pursues ceiling preemption: it uses federal authority to prevent state regulation rather than to create new federal requirements. This conflicts with the floor-preemption design of the AI LEAD Act, which preserves states' ability to adopt stronger protections (Executive Order 14365 — Ensuring a National Policy Framework for AI). The order is analyzed alongside other US approaches in US AI Regulatory Approaches Compared and in the state–federal framework of Techno-Federalism.

A different preemption design was proposed as an alternative in August 2026. Americans for Responsible Innovation's federal frontier-AI blueprint would operate by compliance equivalence rather than displacement: federal compliance satisfies a state requirement only where the federal obligation covers the same conduct and is at least as protective, other state laws remain untouched, states remain free to legislate above the federal floor, and compliance creates no immunity, safe harbor, or presumption against liability under otherwise applicable state law. Where this order seeks preemption of state laws that conflict with its federal policy or burden interstate commerce, the compliance-equivalence model preempts only to the extent of a matched and equally protective federal duty, so the scope of displacement depends on what the federal regime actually requires.

Implementation

Litigation Task Force and the Colorado AI Act

The AI Litigation Task Force was announced on January 9, 2026, one day before its deadline, with a sole mandate to challenge state AI laws inconsistent with the order's policy (Source: whitecase.com).

The order's first operational test came through the Colorado AI Act (SB 24-205), the law it had singled out. The challenge arrived as private litigation that the federal government then joined, not as a suit initiated by the Task Force. xAI filed xAI Corp. v. Weiser (D. Colo. No. 1:26-cv-01515) on April 9, 2026, the first major private-sector constitutional challenge to a state AI law filed after the order (xAI LLC v. Weiser — Complaint (D. Colo. 1:26-cv-01515)). It was not a Task Force action, but it aligned with the order's preemption posture: the order had named Colorado's law as a target of federal concern, and xAI's complaint advanced much the same argument through private enforcement. The complaint pleads six constitutional counts (three First Amendment counts, a dormant Commerce Clause count, a Due Process vagueness count, and an Equal Protection count); a preemption claim based on EO 14365 is not reported among them (xAI LLC v. Weiser — Complaint (D. Colo. 1:26-cv-01515)).

On April 24, 2026 the Department of Justice moved to intervene on the plaintiff's side, the first time the DOJ had intervened in a case challenging a state AI law, arguing that the Act violated the Equal Protection Clause of the Fourteenth Amendment (Source: axios.com). The district court stayed enforcement of the Act on April 27, 2026, and on May 14, 2026 the Colorado legislature passed and Governor Polis signed SB 26-189, repealing and reenacting the 2024 Act as a narrower disclosure-and-transparency framework (Source: nortonrosefulbright.com). The order's named target was thus substantially rewritten before it took effect, through private litigation, DOJ intervention, and the Colorado legislature's replacement of the Act, rather than through the order's own funding-condition and agency-rulemaking mechanisms.

Commerce evaluation and BEAD funding

The Commerce Secretary's § 4 evaluation of onerous state AI laws and the § 5(a) BEAD Policy Notice were both due within 90 days, by approximately March 11, 2026; early tracking by outside counsel noted that the evaluation was not publicly released on that timeline (Source: hklaw.com).

Inside AI Policy reported on July 6, 2026 that NTIA Director Arielle Roth said guidance was coming in summer 2026 on state uses of the $21 billion remaining in the BEAD program, an issue tied to the order's provision making states with "onerous AI laws" ineligible for non-deployment funds, and that congressional Democrats had warned the administration against imposing AI-based funding restrictions on states (Source: insideaipolicy.com).

FTC policy statement

On July 1, 2026 the Federal Trade Commission released for public comment a proposed policy statement responding to the § 7 directive. The proposed statement describes how AI companies that distort their systems' outputs to achieve undisclosed ideological objectives could be deceiving consumers in violation of Section 5 of the FTC Act. The Commission's announcement said that Colorado's Artificial Intelligence Act "appears to coerce companies into altering the output of their AI models" to advance the state's ideological objectives, and that the proposed statement treats such a law as "impliedly preempted to the extent it conflicts with a federal regulatory scheme." Chairman Andrew N. Ferguson said the input would help the Commission formulate a final policy. The Commission voted 2-0 to authorize the Federal Register notice, and comments were due by July 31, 2026 (Source: ftc.gov). Inside AI Policy described the statement as claiming preemption authority over "ideological" state AI laws (Source: insideaipolicy.com). It was the first published agency policy statement under the order, and the first implementation step outside the DOJ litigation track.

Legislative recommendation

In March 2026 the White House released a National Policy Framework for AI, the legislative-recommendation deliverable directed by § 8, setting out recommendations to Congress for a uniform federal approach that would occupy the field the order seeks to clear of state law (Source: hklaw.com).

Legislative follow-on

TRUMP AMERICA AI Act (Blackburn)

In a January 27, 2026 op-ed in The Daily Signal, Sen. Marsha Blackburn (R-TN) presented the framework for her TRUMP AMERICA AI Act as the legislative vehicle that would convert the order's preemption posture into federal law (Why America Needs One Rulebook for AI — Sen. Marsha Blackburn (Daily Signal op-ed, January 2026); Source: dailysignal.com). Blackburn had led the 99-1 Senate vote in 2025 to strip the 10-year state AI moratorium from the reconciliation bill. In the op-ed she contrasts that moratorium, which she says "would eliminate state protections without federal laws to replace them," with the order, which she describes as calling on Congress to pass federal standards, and says her bill "would preempt state protections with strong federal guardrails." The piece is a statement of position in support of the order, not independent evidence of its effects. The framework is organized around what Blackburn calls the "four Cs":

• Children: a duty of care on AI developers to prevent and mitigate foreseeable harm to minors, bundled with the Kids Online Safety Act.
• Creators: protection of voice and likeness, with Tennessee's ELVIS Act held up as the model; bundles the NO FAKES Act and creates a federal right to sue companies for using data and copyrighted material for AI training without "explicit consent."
• Conservatives: regular bias evaluations for "high-risk AI systems," treating political affiliation as a protected characteristic.
• Communities: data-center operators responsible for the full cost of energy and water infrastructure, with no impact on ratepayers.

See also TAKE IT DOWN Act.

American Leadership in AI Act (Lieu–Obernolte)

Reps. Ted Lieu (D-CA) and Jay Obernolte (R-CA) introduced the American Leadership in AI Act (H.R. 8516) on April 27, 2026. It was referred to the House Committee on Science, Space, and Technology and six other committees (Source: congress.gov). Nextgov reported that the bill consolidates more than 20 earlier legislative proposals and recommendations of the House Bipartisan Artificial Intelligence Task Force (Source: nextgov.com). Nextgov's report on its introduction does not identify a provision codifying or extending the order's preemption posture (see Open questions).

Relation to earlier executive orders

Section 1 of the order grounds it in Executive Order 14179 of January 23, 2025 ("Removing Barriers to American Leadership in Artificial Intelligence"), stating that pursuant to that order the President "revoked my predecessor's attempt to paralyze this industry" and directed the removal of barriers to United States AI leadership (Executive Order 14365 — Ensuring a National Policy Framework for AI). The formal revocation of the Biden-era omnibus order, EO 14110, was carried out by a separate instrument, EO 14148 ("Initial Rescissions of Harmful Executive Orders and Actions"), signed January 20, 2025 (Source: federalregister.gov). EO 14365 does not itself revoke EO 14110; it is the successor substantive framework, replacing the Biden-era posture with federal preemption of state AI laws. America's AI Action Plan (July 2025) is the companion strategy document.

Reactions

State and local governments

On December 19, 2025, the National Conference of State Legislatures (NCSL), the Council of State Governments, the National League of Cities, the U.S. Conference of Mayors, and the National Association of State Chief Information Officers (NASCIO) issued a joint statement on the order. It said the order overlooks state and local progress on AI, "creates uncertainty around federal funding," and "moves away from repeated calls for partnership," and that by attempting to preempt state laws it "risks weakening protections for Americans" and could lead to costly litigation. The organizations urged the President to reconsider and proposed an Intergovernmental AI Working Group (Source: nascio.org).

Route Fifty's Chris Teale reported on January 26, 2026 that states were bracing for federal preemption, one month after the signing; the article's summary line states that "Observers called for an intergovernmental working group as well as better collaboration." Teale described the order as having drawn "a stinging reaction from state and local government leaders." At a webinar held with the Public Technology Institute, NASCIO executive director Doug Robinson said that mitigating preemption was "top of mind" for the organization going into 2026, that a preferable model would be a collaborative one "perhaps with a federal floor, with the states be able to build on that," and that preemption is "injurious to the states that have made a lot of progress in protecting their consumers." Robinson predicted a "long, drawn-out court battle" over the order that would take years to resolve (Source: route-fifty.com). The article recalled that Congress had earlier failed to pass a 10-year moratorium on state AI regulation as part of the 2025 reconciliation bill; the bipartisan coalition that defeated that moratorium in summer 2025 is described in How MAGA learned to love AI safety — Nicky Woolf (Transformer, October 2025).

Legal commentary

Farahany, writing on the day of signing, argues that the order is constitutionally vulnerable and unlikely to survive in full. She tests each mechanism against established doctrine: the BEAD conditions against South Dakota v. Dole and NFIB v. Sebelius on the Spending Clause, the Task Force's dormant Commerce Clause theory against Pike v. Bruce Church, and the FCC and FTC actions against West Virginia v. EPA. She frames the order as seeking to accomplish through executive authority what the Senate rejected 99-1 in 2025 (Inside My AI Law & Policy Class 25 (Special Edition): The Executive Order That Could Kill State AI Laws (Farahany, December 2025)).

Open questions

• Whether the order's preemption claims survive a Tenth Amendment challenge is unresolved.
• Whether the AI Litigation Task Force pursues every state AI law or selectively targets the most consequential ones has not been disclosed.
• Whether courts accept the First Amendment theory advanced in xAI v. Weiser, that training-data selection, system-prompt drafting, and guardrail design are expressive editorial activity, is unresolved. A ruling for xAI on that theory would limit state AI regulation independently of the order's own mechanisms; a ruling against it would leave more weight on the FCC and FTC actions, funding conditions, and legislative recommendation.
• Whether the American Leadership in AI Act (H.R. 8516) contains preemption provisions that would codify the order's posture, and whether it advances before the 119th Congress ends on January 3, 2027, is not established.
• Whether the Commerce evaluation under § 4 and the FCC proceeding under § 6 have since been issued has not been established.

Relationships

• depends-on: Executive Order 14179 — Removing Barriers to American Leadership in Artificial Intelligence — § 1 grounds the order in EO 14179
• related: Executive Order 14110 (RESCINDED) — Biden omnibus AI EO, revoked January 20, 2025 by EO 14148; EO 14365 is the later substantive framework
• related: America's AI Action Plan — companion strategy document
• related: Colorado AI Act (SB 24-205) — state-law target cited in § 1; replaced by Colorado SB 26-189 (2026 — replaces 2024 Colorado AI Act)
• related: xAI LLC v. Weiser (challenging the Colorado AI Act) and xAI LLC v. Weiser — Complaint (D. Colo. 1:26-cv-01515) — private challenge to the order's named target, joined by the DOJ
• related: California SB 53, New York RAISE Act, California SB 243 — Companion Chatbots — state laws exposed to preemption
• contradicts: AI LEAD Act (S. 2937) — floor preemption against the order's ceiling preemption
• contradicts: California Executive Order N-5-26 (Trusted AI Procurement) — operates through the state-procurement carve-out
• contradicts: Responsible Innovation at the Frontier (ARI, August 2026) — proposes compliance-equivalence preemption against broad displacement
• related: Procurement-Driven AI Governance — via the § 8(b) state-procurement carve-out
• related: Techno-Federalism, California Effect, US AI Regulatory Approaches Compared
• related: Why America Needs One Rulebook for AI — Sen. Marsha Blackburn (Daily Signal op-ed, January 2026), Marsha Blackburn — TRUMP AMERICA AI Act as proposed legislative vehicle
• related: Lieu-Obernolte AI Bill (April 2026) — American Leadership in AI Act
• related: Federal Trade Commission (FTC), Federal Communications Commission (FCC) — agencies directed under §§ 6–7
• related: Trump v. Slaughter — June 29, 2026 decision making FTC commissioners removable at will, shortly before the FTC's § 7 proposed statement
• related: David Sacks — Special Advisor for AI and Crypto, consulted under §§ 3–7 and co-author of the § 8 recommendation
• related: Inside My AI Law & Policy Class 25 (Special Edition): The Executive Order That Could Kill State AI Laws (Farahany, December 2025), California Just Played the One AI Card the Feds Can't Block (Clearwater, March 2026) — legal analysis of the order
• related: Anthropic v. United States (Pentagon ban challenge) — other federal AI-policy litigation

Sources

• Executive Order 14365 — Ensuring a National Policy Framework for AI — primary text
• xAI LLC v. Weiser — Complaint (D. Colo. 1:26-cv-01515) — parallel private challenge to a named state-law target
• Why America Needs One Rulebook for AI — Sen. Marsha Blackburn (Daily Signal op-ed, January 2026)
• Inside My AI Law & Policy Class 25 (Special Edition): The Executive Order That Could Kill State AI Laws (Farahany, December 2025)
• California Just Played the One AI Card the Feds Can't Block (Clearwater, March 2026)
• Responsible Innovation at the Frontier (ARI, August 2026)
• How MAGA learned to love AI safety — Nicky Woolf (Transformer, October 2025)
• Supporting sources are cited inline.