How the Executive Branch Is Reshaping AI Federalism

Federal preemption is dead; long live preemption.

In December 2025, President Trump issued an executive order that signaled a decisive shift not only in artificial intelligence (AI) policy but also in how regulatory authority is to be governed  in complex, fast-moving domains—a dramatic shift whose significance grew in light of the administration’s March 20 follow-up AI framework. While applying to AI, the mechanisms the order employed—executive coordination, conditional funding, and strategic signaling—are not technology-specific. The method employs an emerging model of top-down Federal governance capable of extending beyond AI—one that shapes the allocation of authority not through conventional legislative or preemptive mechanisms, but through executive coordination operating in advance of them. 

Taken together, the mechanisms employed in the executive order suggest not merely a shift in policy, but rather a distinct model of governance—one that structures how authority is exercised rather, than formally reallocating it. In the traditional framework, such shifts occur through congressional action and are subject to the established preemption doctrine. Here, they instead emerge through unilateral executive action, which controls the conditions under which authority is exercised—structuring incentives and constraints that can meaningfully influence how and whether regulatory authority applies in practice.  In short, traditional principles of preemption don’t apply; they are sidestepped in favor of a rubric of greater top-down control.

Three months after the order, the administration recalibrated its approach. Perhaps censured by calls for respect for states’ rights, the March legislative framework and accompanying press release reintroduced a formal role for the states, emphasizing that traditional state police powers—such as consumer protection, fraud prevention, and child safety—would remain intact. The new framework omitted explicit funding threats and reframed federal action as complementary rather than preemptive. On its face, the shift suggests a return to cooperative federalism.

But the change is more semantic than substantive. Rather than abandoning federal primacy, the administration refined it. Whether deliberate or emergent, the result is a model that might be called “managed federalism”—one in which state authority is preserved in theory but constrained in practice through federal coordination and intergovernmental leverage. This doesn’t amount to regulatory preemption, which would require congressional action. Instead, it reflects a subtler form of displacement: State lawmaking on AI policy remains legally available, yet practically discouraged.

This raises a constitutional question extending far beyond AI, and going to the heart of Federalism and states’ rights. When the executive branch unilaterally shapes not only policy outcomes but the allocation of authority itself—between federal and state governments, and among federal branches—what mechanisms preserve the balance that the Constitution assigns to Congress and to the states? In the absence of statutory preemption or Congressional pushback, the answer may lie less in doctrine than in the cumulative effects of executive coordination, which can, over time, reconfigure democratic lawmaking without formally redrawing it. (It also serves the neat trick of avoiding preemption concerns while directly implicating them.)

Is this model of governance emerging around AI an exception to the longstanding federalist balance, or a testing ground for a broader reconfiguration of how authority is coordinated across the modern administrative state?

From Preemption to Managed Federalism

Executive Coordination in Practice: Mechanism and Effects

The December 2025 executive order did not formally preempt state law. Instead, it deployed a set of familiar administrative tools—executive coordination, agency direction, and conditional funding—uniquely, to sidestep the question. The order articulated a national policy opposing “fragmented” and “onerous” state AI regulation, directed agencies to identify and challenge such laws, and linked federal funding—including broadband infrastructure support not associated traditionally with AI regulation—to the regulatory posture of the states.

To be sure, none of these mechanisms are new. The federal government routinely coordinates across agencies and attaches conditions to grant funding. What distinguishes the December 2025 order was how those tools are directed and used together—not to implement a congressional framework, but to shape the regulatory environment in advance of one. In doing so, the order may shift the trajectory of future lawmaking.

Formally, states retain their authority. No statute directly displaced their power to regulate in areas such as consumer protection, education, or fraud. In practice, however, the executive has shifted incentives. Legislatures operating under the prospect of federal challenge or funding consequences now face a different calculus. Legislative proposals that might otherwise have advanced have moved more cautiously or stalled. SeeThis Utah, where an AI bill introduced by state Rep. Doug Fiefia (R)—which would have imposed safety and transparency obligations on AI developers—initially advanced with unanimous committee support but later lost momentum. The bill failed following opposition from federal officials and industry stakeholders. The effect was not prohibition, but rather deterrence.  It ends up in the same place.

The March 2026 framework recalibrated this approach without limiting it. It emphasized traditional state police powers and frames federal policy as establishing a national baseline while allowing room for state action. But this shift in framing was more rhetorical than substantive. The December 2025 executive order operates as a formal directive within the executive branch, establishing concrete mechanisms for agency coordination and evaluation of state law. By contrast, the March framework primarily articulates policy guidance, including references to traditional state police powers, but does not modify or rescind the earlier directive, leaving the underlying structure intact. The framework continues to emphasize avoiding regulatory fragmentation in favor of national uniformity, preserving mechanisms through which federal agencies can evaluate and discourage divergent state approaches—all in the absence of a comprehensive congressional framework.

This is semantics. The shift reflects a move from overt assertions of federal primacy to a more calibrated model that reduces the visibility of coercive tools while maintaining their effect. State authority is technically preserved, but substantively conditioned—consistent with the model of managed federalism described above. We end up in the same place.

A similar dynamic appears beyond federal-state relations. In addressing the use of copyrighted material to train AI systems—an area of copyright law constitutionally assigned to Congress—the March 2026 framework declines to propose legislation and instead directs resolution to the courts. This reflects a broader pattern: Rather than waiting to reallocate authority through legislation and established preemption frameworks, the executive branch proactively  shapes how and where decisions happen.

Managed Federalism and the Allocation of Authority

Presidents act squarely within their traditional role when they engage in agenda-setting advocacy aimed at persuading Congress to enact legislation. More difficult questions arise, however, when executive action shapes the conditions under which lawmaking occurs—particularly in instances where the executive branch operates in advance of legislation and directly influences state regulatory choices. In those circumstances, the issue is not simply influence over policy outcomes, but whether the allocation of authority is being reconfigured outside of the processes the Constitution assigns to Congress.

Of course, the executive branch has long influenced policymaking. Major legislative initiatives—from civil rights enforcement to the Affordable Care Act—underscore the central role of presidential leadership. At times, federal administrations have used funding leverage to influence state policy, particularly in the civil rights context, where the withdrawal of federal funds was tied to compliance with statutory mandates. But those episodes have operated either through Congress or pursuant to existing statutory frameworks, and were directed toward enforcing federal law, with Congress ultimately determining the allocation of authority.

National Federation of Independent Business v. Sebelius, which evaluated the Affordable Care Act, underscores this distinction: The Supreme Court held that conditional funding may become unconstitutional when it effectively compels state participation by threatening the loss of existing funding. That constraint operated within a statutory framework enacted by Congress and subject to judicial review. Similar pressures may arise, however, through executive coordination outside a comprehensive statutory scheme, such as with the recent efforts to manage AI regulation, which move beyond the doctrinal framework through which the courts have traditionally assessed such coercion.

Even if Congress ultimately codifies the White House’s approach, the distinctive feature of the present model is that it shapes state legislative incentives beforehand in ways that may approach the kinds of pressures identified in National Federation of Independent Business v. Sebelius—potentially altering the trajectory of lawmaking before Congress acts at all. The result is a more systematic and anticipatory form of influence, structuring legislative incentives before the fact rather than contesting law afterward.

Implications for Constitutional Doctrine

Comparing this model to more familiar, statute-driven allocations of authority reveals its constitutional significance. In areas such as employment discrimination law, Congress sets national baselines while preserving state authority to expand on them. Federal statutes such as Title VII establish minimum protections without occupying the field, and courts enforce the resulting boundaries. The allocation of authority is explicit and legislatively determined, in contrast to the model of managed federalism employed for AI governance.

This distinction has important implications for constitutional doctrine. Courts are well equipped to address direct conflicts between federal and state law, but less suited to shifts that occur through incentives and expectations. Under the judicial doctrine of standing, only the impacted states can challenge such an executive order, even though all citizens are impacted when their most direct channel to the ideal of self-governance—the states—is closed. When authority influences this way, the issue is no longer simply whether federal action preempts state law, but whether the conditions for meaningful lawmaking remain intact.

AI makes this dynamic particularly visible, but it is far from unique. As emerging technologies implicate national security, economic competition, and interstate systems, similar pressures toward centralized coordination will arise. The mechanisms deployed here are readily transferable. In that context, constraints on state authority may increasingly emerge not through enacted law, but through the structuring of incentives and decision-making processes—producing effects that resemble preemption without invoking the constitutional mechanisms that traditionally authorize it.

State Lawmaking in Practice: A Case Study in Constraint

At first glance, continued growth in state AI legislation might seem to cut against any claim of federal displacement. But in a rapidly expanding domain, legislative volume alone is a poor proxy for regulatory autonomy. The more relevant question is whether certain types of state interventions—particularly those targeting AI developers or frontier systems—are less likely to be proposed, narrowed in scope, or fail to advance following federal intervention. On that measure, sustained activity may coexist with, and even mask, a more selective form of constraint.

The practical effects of this model are best understood in legislative behavior. In late 2025, as AI governance discussions accelerated across multiple states, policymakers explored measures in areas within traditional state authority, including consumer protection, election integrity, and deceptive media practices.

One such effort in Arizona involved a proposed transparency measure addressing AI-generated or materially altered media. The proposal would have required disclosure when synthetic content was presented in a way likely to mislead a reasonable viewer—a thoughtful approach grounded in fraud prevention and consumer protection. At the time, no federal framework addressed this risk, and the proposal fell squarely within traditional state domains.

Prior to the December 2025 executive order, the legislature had begun to engage with the proposal, based on this article’s authors’ involvement in early-stage discussions. Initial outreach generated sufficient interest to invite further development.

After the order, that momentum halted—a result that may reflect, at least in part, a changed policy environment in which the direction of federal policy had become less certain and the costs of divergence more salient.

While early-stage, this example illustrates the logic of managed federalism. The executive branch need not invalidate state law to control it. Signaling that certain regulatory approaches are disfavored and placing previously approved federal funding in jeopardy alters legislative incentives, as reported recently in Missouri and in Louisiana. As do private lawsuits challenging the constitutionality of state efforts to pass regulations on AI, as filed by x.AI against Colorado’s soon-to-take-effect AI law; x.AI’s suit cites the December 2025 executive order and March 2026 AI Framework and was just joined by the Department of Justice. The result is not necessarily reversal, but attenuation—initiatives stall before fully taking shape. In compressed legislative timelines, hesitation at early stages can determine whether proposals advance.

This dynamic does not formally eliminate state authority. The March framework reaffirms traditional police powers, and no statute currently displaces state authority in these areas. But the distinction between formal authority and practical capacity is consequential. A legislature that hesitates amid federal uncertainty operates under different constraints than one responding solely to local conditions.

This gap is where the constitutional stakes become visible. Federalism depends not only on the formal allocation of power, but on the practical capacity to exercise that power. When executive action shapes those conditions, the balance between national coordination and state autonomy can shift without a corresponding change in law.

The Allocation of Authority in the Age of AI

The developments described here point to a structural shift: not only how AI will be governed, but how authority over that governance is being shaped.

The emerging model operates not through formal preemption, but through coordination—structuring incentives and institutional roles in ways that influence who acts and under what constraints. In that environment, authority can shift without being openly reassigned.

That shift carries constitutional significance. The system allocates authority through identifiable mechanisms: Congress determines when national law displaces state authority, states retain primary regulatory power absent such displacement, and courts resolve conflicts between the two. When those allocations are altered indirectly—through executive action shaping decision-making itself—the usual pathways for correction become less certain. While executive influence over state policymaking is not new, the model described here operates more systematically and in advance of legislation, shaping state regulatory incentives across an entire domain rather than through episodic interventions tied to specific statutory frameworks.

This dynamic is not unique to AI, but AI makes it visible. As similar pressures arise elsewhere, the central question will not be whether some degree of coordination is necessary, but whether the allocation of authority remains anchored in constitutional processes. Traditionally, the reallocation of state authority occurs through legislation and is mediated by courts applying the preemption doctrine. The model described here suggests that similar effects may arise without those mechanisms.

The result is not simply more centralized governance. It is a quieter shift: from decisions made through law to decisions shaped through control over how lawmaking itself is structured and exercised. Whether existing institutions are equipped to respond may shape how AI is governed. More fundamentally, the question is whether the Constitution’s allocation of authority remains operative in practice—or whether it is gradually supplanted by control over the processes that define it.