Preemption by Other Means: Managed Federalism and US AI Rules

Preemption by Other Means: How "Managed Federalism" Is Quietly Reshaping Who Sets AI Rules

A weak signal in regulation and policy: beneath the loud preemption debate, the federal executive is using conditional broadband funding, a DOJ litigation task force and agency signalling to chill state AI lawmaking in practice without preempting it in law, producing a deterrence-shaped patchwork on a toolkit portable to other domains.

The consensus on US AI rules in 2026 is that the country is heading either to a Congressional national standard or to a fifty-state patchwork held together by the courts. The non-obvious signal is that neither outcome is what is happening. The 11 December 2025 executive order on AI does not preempt state law; it deploys conditional broadband funding, a DOJ litigation task force and aligned FTC and FCC action to deter, narrow or stall state AI legislation before it lands. Lawfare's specialist legal analysts call this "managed federalism" (Lawfare, May 2026). The durable 2026-2028 outcome is a deterrence-shaped, contested patchwork, not a clean national rule, and the same toolkit is portable.

Signal Identification

The shift is institutional, not doctrinal. The visible debate is whether an executive order can lawfully preempt state AI statutes. The non-obvious move is that the administration has routed around that question: not preempting state authority on paper while constraining it in practice through funding leverage and federal litigation. The change relocates rule-setting toward executive coordination and reshapes the operating environment for any company deploying AI across US states.

Time horizon: 2-3 years (December 2025 EO mechanisms operationalising through 2026; BEAD ineligibility determinations and DOJ task-force litigation set 2026-2028). Plausibility band: Medium-High. Geographic / Jurisdictional Scope: Primary: the United States, specifically the federal-state allocation of AI regulatory authority. Spillover: companies operating across US states, and other contested policy domains open to the same coordination-and-leverage toolkit. Sectors exposed: AI developers and deployers, regulated industries using AI (financial services, insurance, employment, housing, healthcare), state attorneys general, state legislators, BEAD-recipient telecoms, governance and compliance functions.

What's Changing

The 11 December 2025 executive order does not itself preempt state law. It directs the DOJ to set up an AI Litigation Task Force, instructs Commerce to evaluate "onerous" state AI laws and to make states with such laws ineligible for remaining Broadband Equity, Access, and Deployment (BEAD) funds, lets agencies condition discretionary grants, and pulls the FTC and FCC into the package (The White House, December 2025). A uniform national standard would still require Congress.

Brookings is blunt: "The order itself has no preemptive power whatsoever." It works because it is "designed to intimidate" through DOJ litigation, funding threats and FTC and FCC action, while the March 2026 framework is "a vague and abstract roadmap" rather than enforceable standards (Brookings Institution, April 2026). Lawfare captures the structural point: the order deploys executive coordination, conditional funding and litigation to sidestep the preemption question, with state authority preserved in theory but constrained in practice; bills in Utah, Missouri and Louisiana have stalled or been narrowed (Lawfare, May 2026).

The litigation track moved from theory to practice on 24 April 2026, when the DOJ intervened in xAI's suit against Colorado's AI Act (SB24-205) on Equal Protection grounds, the first time the federal government has moved to invalidate a state AI law (US Department of Justice, April 2026). The Act is now effectively frozen weeks before its 30 June 2026 effective date: a magistrate stayed enforcement on 27 April, the Colorado Attorney General has paused rulemaking, and a replacement narrowing scope and pushing the effective date to 1 January 2027 is under negotiation (The National Law Review, May 2026). The recalibrated 20 March 2026 National AI Legislative Framework concedes the executive cannot do this alone, asking Congress to legislate because "a patchwork of conflicting state laws would undermine American innovation" (The White House, March 2026).

The managed-federalism toolkit: how state AI lawmaking gets chilled without being preempted

Federal levers converge on state AI bills without formal preemption (Lawfare, May 2026; directional).

Disruption Pathway

The pathway runs in three stages. First, signalling: the December 2025 order publicly threatens BEAD ineligibility, DOJ litigation and agency action, and Commerce begins compiling its evaluation of "onerous" state laws. The threat alone reprices the legislative calculus in state capitols. Second, demonstration: the DOJ intervenes against Colorado's AI Act, enforcement is stayed, and rulemaking pauses pending a narrower replacement. The package shows other states what active resistance costs. Third, attenuation: bills stall or narrow in Utah, Missouri and Louisiana before the federal government contests a single statute on the merits (Lawfare, May 2026).

Stress concentrates at three points. Funding dependence: BEAD-reliant states face an asymmetric trade between holding an AI rulebook and securing broadband money already planned. Litigation exposure: any state pushing through a contested AI statute now faces a federal Equal Protection challenge as a first move, not a last. Coordination cost for business: the patchwork that emerges is shaped by what survives federal deterrence, more uneven and less predictable than either a clean national standard or transparent state-level fragmentation. KPMG flags that regulatory fragmentation has shifted "from anomaly to strategic focus", forcing companies to run multiple risk classifications and model-governance processes (KPMG, March 2026).

Why This Matters

For boards, general counsel and policy teams, the assumption that needs revising is that the 2026-2028 US AI rulebook will be settled in Congress or the courts. On the evidence, the operative venue is executive coordination: which state AI rules survive will be decided more by federal funding leverage and DOJ litigation choices than by state legislative majorities or constitutional doctrine. Compliance scanning of statutes on the books will mis-price the real rule set; the real rule set is what federal deterrence leaves standing. State lawmakers face a quantified federal cost for legislative ambition, and effective lobbying has shifted toward the executive branch. The next two cycles set the template for whether this toolkit becomes the default mode of contested policymaking.

Decision-action posture for this signal: Prepare. The mechanism is operational and the first demonstration case has landed; map BEAD-state exposure and litigation-track risk now and ready governance for a non-uniform end state, without abandoning state-by-state compliance.

Counter-Argument

The strongest objection is that this reads too much into an executive instrument with limited legal force, and underweights the lane that actually decides US AI rules: Congress. The administration concedes the point, asking Congress to legislate because the framework "can succeed only if it is applied uniformly" (The White House, March 2026). Brookings stresses that the order has "no preemptive power whatsoever" and that BEAD and DOJ pathways "each face serious legal obstacles" (Brookings Institution, April 2026). On this view, managed federalism is a holding pattern that ends when Congress legislates or courts strike down the operative theories.

The objection is real but partial. Even if Congress eventually legislates, the toolkit has already chilled state lawmaking through 2026 and into 2027, the period in which much of the contested rulebook would otherwise have been written. Even if courts ultimately narrow BEAD-conditioning and the DOJ theory, the time cost to states and the de facto template outlast the doctrinal outcome.

Implications

Taken together, the sources point to a durable institutional shift in how contested federal-state disputes are managed, not a one-off AI episode. The 2026-2028 inflection is whether managed federalism stabilises as a recognised mode of executive coordination, portable to other domains, or is rolled back before it sets that template. Winners build a compliance posture for a deterrence-shaped patchwork and engage at the executive-coordination layer, not only in state capitols. The contest is shifting from who writes AI rules to which AI rules survive federal pressure.

Early Indicators to Monitor

Commerce publishes its evaluation of "onerous" state AI laws and BEAD ineligibility determinations follow.
The DOJ Litigation Task Force opens a second intervention against a non-Colorado state AI statute.
Additional state AI bills stall or narrow after explicit federal funding or litigation signalling.
Colorado's legislative replacement passes (or fails, returning the original Act to the docket).
A state attorney general or coalition files a counter-suit challenging BEAD or grant conditioning as unconstitutional coercion.

Disconfirming Signals

Congress enacts a national AI law that genuinely preempts state statutes, retiring the managed-federalism workaround.
A federal court strikes down BEAD-conditioning or the DOJ Equal Protection theory.
State AI lawmaking accelerates rather than slows through late 2026, with new statutes passing outside BEAD-dependent states.
The DOJ withdraws from the Colorado intervention or loses on the merits early.
The administration formally abandons the executive-coordination route for legislative negotiation.

Strategic Questions

Should AI deployers build compliance for the statutes on the books, or for the rules federal deterrence is likely to leave standing?
At what point is the managed-federalism toolkit normalised enough to deploy in non-AI contested domains?
Do state legislators continue to legislate against federal pressure, or quietly retire bills?

Keywords

Managed federalism; AI regulation; preemption; executive order; BEAD funding; DOJ litigation task force; Colorado AI Act; SB24-205; xAI; National AI Legislative Framework; regulatory fragmentation; state attorneys general; FTC; FCC; federalism

Bibliography

Source tiers: Tier 1, governments, regulators and intergovernmental bodies. Tier 2, think-tanks, academic institutes, major consultancies and quality data providers. Tier 3, quality journalism and specialist trade press. Tier 4, vendor, company and practitioner sources, used only as directional corroboration.

Tier 1 Ensuring a National Policy Framework for Artificial Intelligence (Executive Order). The White House (11/12/2025).
Tier 1 President Donald J. Trump Unveils National AI Legislative Framework. The White House (20/03/2026).
Tier 1 Justice Department Intervenes in xAI lawsuit Challenging Colorado's "Algorithmic Discrimination" Law. US Department of Justice (24/04/2026).
Tier 2 States can and should regulate AI in criminal justice. Brookings Institution (16/04/2026).
Tier 2 Regulatory Fragmentation: From Anomaly to Strategic Focus. KPMG (18/03/2026).
Tier 3 How the Executive Branch Is Reshaping AI Federalism. Lawfare (04/05/2026).
Tier 3 The Colorado AI Act Hits a Wall: Litigation, Legislative Uncertainty, and an Enforcement Standstill. The National Law Review (01/05/2026).

Analyst inferences and editorial framing

Claim-fidelity self-disclosure. The "managed federalism" framing and the attenuation-of-state-lawmaking pathway are taken verbatim and as faithful summary from Lawfare (04/05/2026). The "no preemptive power whatsoever" and "designed to intimidate" characterisations are verbatim from Brookings (16/04/2026). The "patchwork of conflicting state laws would undermine American innovation" line is verbatim from the 20 March 2026 White House framework. The DOJ-intervention, xAI-suit and Colorado-stay facts are faithful summary of the DOJ press release (24/04/2026) and the National Law Review analysis (01/05/2026). The "deterrence-shaped, contested patchwork" framing is analyst characterisation, signposted in the Implications and Why This Matters sections. The diagram is a directional schematic; the four-lever framing is analyst synthesis of the EO's operative provisions as catalogued in the White House and Lawfare sources.