AI Policy Wiki
Dashboard

Inside My AI Law & Policy Class 25 (Special Edition): The Executive Order That Could Kill State AI Laws (Farahany, December 2025)

medium confidence · updated 2026-06-06

Special edition class on the December 11 2025 Trump federal preemption EO ('Ensuring A National Policy Framework for Artificial Intelligence'). Walks through the EO's 5 mechanisms (evaluation/designation, BEAD funding conditions, DOJ Litigation Task Force, FCC/FTC rulemaking, legislative recommendations). Tests each against constitutional doctrine — South Dakota v. Dole, NFIB v. Sebelius, Pike v. Bruce Church, West Virginia v. EPA. Notes Senate's 99-1 vote against the same outcome via legislation.

Author: Nita Farahany Source: https://nitafarahany.substack.com/p/the-executive-order-that-could-kill Published: December 11, 2025

A special-edition installment (Class 25b of 27) in Nita Farahany's AI Law & Policy course, published December 11, 2025, analyzing the Trump administration's federal preemption executive order, "Ensuring A National Policy Framework for Artificial Intelligence" (Executive Order 14365 — Ensuring a National Policy Framework for AI). The class frames its inquiry around the question of whether the United States can compete globally with fifty separate state AI regulatory regimes, walks through the EO's five mechanisms, and tests each against established constitutional doctrine. Farahany argues the order is constitutionally vulnerable and unlikely to survive in full.

Summary of argument

The class situates the EO against the preceding legislative cycle. In 2025, more than 1,000 AI-related state bills were introduced and more than 75 became law. Congress had attempted a federal response: the "One Big Beautiful Bill" included a 10-year state AI moratorium tied to BEAD funding, but the Senate voted 99-1 to remove it, with bipartisan opposition that Farahany attributes in part to Senator Marsha Blackburn (R-TN) leading Republican resistance. Farahany's framing is that the EO seeks to accomplish through executive authority what Congress overwhelmingly rejected.

The five mechanisms and their constitutional vulnerabilities

Farahany works through each of the EO's five mechanisms and pairs each with a constitutional doctrine she argues it is exposed to.

Evaluation and designation process. Within 90 days, Commerce is to identify state laws that require AI to "alter their truthful outputs," compel disclosure violating the First Amendment, or conflict with "minimally burdensome" policy. Farahany notes the EO defines neither "AI" nor "onerous," and that David Sacks (Special Advisor for AI and Crypto) holds significant influence over the process while being required to consult other officials. She characterizes the framing of anti-discrimination requirements as forcing "false results" as a polemical move.

Federal funding conditions. States with laws designated "onerous" would lose access to BEAD Program funds, which Farahany identifies as $42.45 billion for rural broadband. She tests this against the Spending Clause line of cases: South Dakota v. Dole (1987) permits conditional spending where the condition relates to a federal interest, while NFIB v. Sebelius (2012) warns that making existing federal funds contingent on new policy can amount to coercion rather than inducement. She observes that the Infrastructure Act said nothing about AI.

DOJ Litigation Task Force. The Attorney General is to establish a task force within 30 days to challenge state laws on Dormant Commerce Clause, federal preemption, and other grounds. Farahany measures this against Pike v. Bruce Church (1970), under which non-discriminatory state regulations stand unless their burdens are "clearly excessive" relative to local benefits, noting that most state AI laws apply equally to in-state and out-of-state actors.

Agency rulemaking. The FCC is to consider federal disclosure standards that would preempt state laws, and the FTC is to issue a policy statement on when state laws conflict with deceptive-practices prohibitions. Farahany invokes West Virginia v. EPA (2022) — under which agencies cannot make major policy decisions without clear congressional authorization — and observes that neither the FTC Act nor the FCC's enabling statutes mention AI.

Legislative recommendations. The EO directs the development of legislation to preempt state AI laws while exempting child safety, AI compute and data center infrastructure (except permitting reforms), state government AI use, and others to be determined. Farahany emphasizes that the child-safety exemption applies only to the legislative-recommendations mechanism, not to the enforcement mechanisms (DOJ litigation, BEAD funding loss, agency rulemaking). On that reading, laws such as the NY SAFE Act and the California Age Appropriate Design Code could still face DOJ litigation. She offers three interpretations of the partial exemption: oversight (poor drafting structure), flexibility (case-by-case treatment as sensible), and priority reveal (that genuinely paramount child-safety concerns would have been exempted from all five mechanisms).

Broader analysis

Drawing on a point she attributes to Randall Cook, Farahany argues that foreign capital — citing the UAE's commitment of $25 billion to US AI data centers — requires regulatory certainty, and that America's competitive advantage should come from coherence grounded in democratic legitimacy: stable frameworks that survive political transitions. On her account the EO supplies neither, instead exercising uncertain federal power through discretionary authority that a subsequent administration could reverse.

She frames three principles she sees in tension across the dispute: state experimentation versus federal coordination, democratic process (the Senate's 99-1 vote) versus executive urgency, and innovation versus safety. On timing, she notes that even if courts rule against parts of the EO, such litigation takes years, while the 2026 state legislative sessions are already under way and constitutional clarity will come later.

Provenance

Published on Farahany's Substack on December 11, 2025 (Source: https://nitafarahany.substack.com/p/the-executive-order-that-could-kill), the same day as the EO. As a class session it presents Farahany's own legal analysis and reflects positions argued rather than settled outcomes.

Relationships