AI Preemption War: Part 1 — The Federalism Question
Series: The AI Preemption War | Table42 Research
> Note on scope: The White House’s March 2026 National Policy Framework for Artificial Intelligence is a non-binding policy document, not legislation. This series analyzes its implications for potential federal action and the broader preemption debate. Where we discuss what courts “will face” or what preemption “would erase,” we are projecting from policy direction to legislative possibility—not describing current law.
—
The Constitution does not answer the question plainly. Commerce Clause gives Congress power to regulate interstate economic activity, but where AI begins and ends, and what states can do in the spaces Congress hasn’t occupied, remains contested terrain. The White House’s March 2026 National Policy Framework for Artificial Intelligence has pushed that contest to center stage—and Congress now faces a decision that will define the architecture of American AI governance for decades.
The Preemption Problem
Federal preemption of state law is not itself unconstitutional. Congress can displace state regulation when it chooses to, and courts generally defer when Congress speaks clearly. The Supreme Court’s Buckman v. Plaintiffs’ Legal Committee (2001) upheld federal occupation of medical device safety against state tort claims; Arizona v. United States (2012) affirmed that where Congress has enacted a comprehensive regulatory scheme, states cannot enact laws that stand as obstacles to federal objectives. Preemption is a legitimate tool.
But preemption is not unlimited. The Constitution’s anti-commandeering doctrine—New York v. United States (1992), Printz v. United States (1997)—prevents the federal government from compelling states to enforce federal priorities. States retain authority to act in domains Congress hasn’t occupied. And where preemption language is ambiguous, courts apply the presumption against preemption in areas of traditional state authority—health, safety, welfare. Police powers remain with the states unless clearly displaced.
The AI question sits at the intersection of these principles. States have been regulating AI for years: California’s transparency requirements, Colorado’s algorithmic discrimination rules, Illinois’ employment disclosure statutes, Texas’s innovation sandbox. These laws rest on state police powers—authority to protect consumers, workers, and residents from fraud, discrimination, and harm. Congress hasn’t occupied this field. The question is whether it now chooses to.
The Constitutional Architecture
Three analytical frames matter here.
The “Field Preemption” Question. Congress can occupy a regulatory field so completely that state laws are displaced even without explicit preemption language. Under Gade v. National Solid Wastes Mgmt. Ass’n (1992), when federal regulation is “so pervasive” that Congress left no room for states to act, field preemption applies. If Congress passes a comprehensive AI framework with detailed federal standards, it may implicitly displace state laws even in areas traditionally reserved to the states. Courts will look at congressional intent, regulatory density, and the balance of federal-state interests.
The “Conflict Preemption” Question. Even without field preemption, state laws that conflict with federal objectives are void. Under Cipollone v. Liggett Group, Inc. (1992), state law is preempted when it “stands as an obstacle to the accomplishment and execution of the full purposes and objectives of Congress.” If the federal framework aims for national uniformity—explicitly—that aim creates conflict with state-level variation.
The “Burden/Benefit” Balance. The framework invokes a cost-benefit logic: state laws that are “unduly burdensome” to AI development can be preempted. This framing matters. Courts have sometimes accepted proportionality reasoning in preemption analysis—the argument that compliance costs justify displacement. But the “burden” here is framed in terms of industry compliance costs, not public welfare. The question is whether that framing carries legal weight or is just policy rhetoric.
The Constitutional Floor Problem
Here’s where the architecture gets unstable.
The framework preserves state authority over “generally applicable laws,” law enforcement, and zoning. This sounds like a carve-out for state police powers. But “generally applicable” is undefined, and “burdensome” is doing significant work. A state employment discrimination law applies generally to all employers—but it imposes specific costs on AI-enabled hiring systems. Is it preempted? The framework doesn’t say.
Compare this to other federal preemption regimes. The National Labor Relations Act implicitly preempts state laws that conflict with collective bargaining policy—but preserves state jurisdiction over purely local matters. The Employee Retirement Income Security Act preempts state insurance regulation in specific domains—but not general contract law. In each case, Congress drew lines that courts have refined over decades.
For AI, Congress hasn’t drawn those lines yet. The framework is a statement of intent, not a statute. Its preemption language has no legal effect until Congress passes a law. The real question is what that law looks like—and whether it will be written to preempt broadly or to carve out protected state space.
The Constitutional Question the Courts Will Face
If Congress adopts the framework’s approach, courts will face the same questions they’ve faced in other preemption contexts:
1. Does Congress have the authority? Commerce Clause covers activities with substantial interstate effects. AI development and deployment is arguably interstate by nature—models trained on global data, deployed across state lines, affecting commerce in every state. But Congress’s power is not unlimited. If a federal AI law reaches purely local AI applications with no meaningful interstate connection, courts may find the mandate exceeds Congress’s authority.
2. Is preemption explicit enough? Under Rice v. Santa Fe Elevator Corp. (1947), courts presume against preemption in areas of traditional state authority. The more explicit Congress is about intending to displace state law, the more likely courts will sustain preemption. If the statute is vague about which state laws are displaced, courts will read the ambiguity in favor of the states.
3. Does preemption conflict with constitutional structure? The anti-commandeering doctrine doesn’t prevent Congress from preempting state law—but it does prevent Congress from compelling states to enforce federal priorities. If the framework tries to condition federal AI funding on state compliance with federal standards, that raises different constitutional questions than straightforward preemption.
The Case for Preemption
Before evaluating what preemption would displace, the strongest version of the argument for federal preemption deserves honest presentation.
National market uniformity. AI development and deployment are inherently interstate. Models trained on global data serve customers across all 50 states. A patchwork of state requirements imposes real compliance costs—not just on large companies, but on startups and small businesses with no legal departments navigating contradictory rules across jurisdictions. The dormant Commerce Clause exists partly to prevent states from enacting regulations that burden interstate commerce. State AI laws that impose disparate requirements on companies serving a national market raise legitimate Commerce Clause concerns, even without federal preemption (Pike v. Bruce Church, Inc., 397 U.S. 137 (1970)).
Regulatory capture at the state level. The series’ framing assumes that state regulation is more democratically accountable than federal regulation. Public choice theory suggests the opposite: smaller jurisdictions are often more susceptible to regulatory capture because the costs of organizing resistance are higher relative to the benefits of capture. State-level lobbying is cheaper and less scrutinized than federal lobbying. The assumption that state AI laws represent “democratic choices” rather than industry-influenced compromises deserves the same scrutiny applied to federal policy.
International competitiveness. The EU and China have adopted national AI strategies. The US lacks a comparable national framework, creating uncertainty for companies developing AI for international markets. A federal floor—even a permissive one—establishes a baseline that international partners can engage with and that companies can plan around. State fragmentation undermines the US negotiating position in international AI governance discussions.
Historical precedent for federal consolidation. Most major regulatory domains—labor relations, securities, environmental protection, pharmaceutical safety—eventually consolidated at the federal level after periods of state experimentation. The NLRA, ERISA, and the Clean Air Act all preempted state regulation to establish national standards while preserving some state authority. This pattern is neither unusual nor inherently hostile to federalism; it reflects the practical reality that some economic activities require national rules.
These arguments don’t settle the question. But they establish that the preemption debate has genuine constitutional and structural foundations—not merely industry preference. The question is whether preemption in this specific case produces better outcomes than the alternative.
The Federalism Stakes
What’s actually at issue here is not merely jurisdictional—it’s about where accountability lives.
States regulate AI because constituents demand it. California, Colorado, and Illinois passed AI laws because voters wanted protections against algorithmic discrimination, transparency in automated decisions, and safeguards for workers displaced by automation. Those laws are imperfect, sometimes overbroad, occasionally contradictory across states. But they reflect democratic choices made at the state level, accountable to state voters.
Federal preemption displaces those choices with national uniformity. That has benefits—reducing compliance fragmentation, establishing clear national standards, preventing a race to the bottom where states compete to attract AI investment by weakening protections. But it also centralizes AI governance in a federal system where AI companies have substantial influence, and where Congress moves slowly and inconsistently.
The question isn’t whether federal preemption is constitutional. It’s whether centralized federal control over AI governance serves the same accountability purposes as state-level regulation—or whether it just shifts the locus of influence from state legislators to federal administrators and the industry players who shape them.
Primary Sources
Federal Framework
1. Executive Order 14365 (December 11, 2025): “Ensuring a National Policy Framework for Artificial Intelligence” — foundation for the March 2026 framework
2. National Policy Framework for Artificial Intelligence (March 20, 2026): White House legislative recommendations — four-page framework with seven pillars
3. Senator Marsha Blackburn’s TRUMP AMERICA AI Act (discussion draft, March 2026): 291-page alternative that pairs preemption with detailed federal standards
Supreme Court Preemption Precedent
4. New York v. United States, 505 U.S. 144 (1992): Anti-commandeering doctrine establishing federal limits on compelling state action
5. Printz v. United States, 521 U.S. 898 (1997): Extended anti-commandeering to executive branch officials
6. Buckman v. Plaintiffs’ Legal Committee, 531 U.S. 341 (2001): Field preemption of state tort claims in federally regulated domains
7. Cipollone v. Liggett Group, Inc., 505 U.S. 504 (1992): Conflict preemption framework for state laws that “stand as an obstacle”
8. Gade v. National Solid Wastes Management Ass’n, 505 U.S. 88 (1992): Field preemption when federal regulation is “so pervasive” that states have no room to act
9. Rice v. Santa Fe Elevator Corp., 331 U.S. 218 (1947): Presumption against preemption in areas of traditional state authority
10. Arizona v. United States, 567 U.S. 387 (2012): Federal immigration scheme preempts conflicting state laws; comprehensive federal regulation displaces state action
Federal Statutory Frameworks
11. National Labor Relations Act, 29 U.S.C. § 151 et seq.: Preempts state laws conflicting with collective bargaining policy while preserving local jurisdiction
12. Employee Retirement Income Security Act, 29 U.S.C. § 1001 et seq.: Preempts state insurance regulation in specific domains while preserving general contract law
State AI Laws
13. California AB 2013 (Chapter 673, Stats. 2024): Generative AI transparency requirements — signed September 28, 2024, effective January 1, 2026
14. Colorado SB 24-205 (2024): Colorado Artificial Intelligence Act / Anti-Discrimination in AI Law (ADAI) — first comprehensive state AI consumer protection law, effective February 1, 2026
15. Illinois HB 3773 (Public Act 103-0606, 2024): Amends Illinois Human Rights Act to require employer notice when AI influences employment decisions, effective January 1, 2026
16. Texas HB 149 (TRAIGA, 2025): Texas Responsible Artificial Intelligence Governance Act — comprehensive AI governance with prohibited uses, disclosure requirements, and civil penalties, signed June 22, 2025
17. New York Local Law 144 (2023): Automated Employment Decision Tool (AEDT) bias audit requirements
18. Washington HB 2031 (2023): Synthetic media and deepfake disclosure requirements
19. Maryland HB 276 (2022): Facial recognition use limitations for law enforcement
20. Virginia HB 2034 (2022): Consumer data protection with AI provisions
Additional Supreme Court Precedent
21. Murphy v. NCAA, 584 U.S. 453 (2018): Anti-commandeering applies to affirmative prohibition on state legalization
22. Arizona v. United States, 567 U.S. 387 (2012): Field preemption analysis in immigration context
23. Bates v. United States, 522 U.S. 23 (1997): Express preemption analysis methodology
24. Crosby v. National Foreign Trade Council, 530 U.S. 363 (2000): Conflict preemption where state law obstructs federal objectives
25. Gade v. National Solid Wastes Mgmt. Ass’n, 505 U.S. 88 (1992): Field preemption when federal regulation is “so pervasive”
Federal Statutory Frameworks
26. Clean Air Act § 116, 42 U.S.C. § 7416: Federal floor with state opt-up authority — model for AI
27. Clean Water Act § 510, 33 U.S.C. § 1370: State authority to adopt more stringent standards
28. National Environmental Policy Act (NEPA), 42 U.S.C. § 4321 et seq.: Federal-state coordination model
29. Occupational Safety and Health Act, 29 U.S.C. § 667: State plans with federal approval
30. Health Insurance Portability and Accountability Act (HIPAA), 42 U.S.C. § 1320d et seq.: Federal floor with state opt-up for privacy
31. Fair Credit Reporting Act (FCRA), 15 U.S.C. § 1681 et seq.: Preemption with state exceptions
—
Status: Draft | Citation count: 31 | Next: Part 2 — The Laboratory States