The Obstruction Doctrine: What Level of Governance Is Appropriate for AI?

Kevin T. Frazier

The proper delineation of authority between the federal government and the states has often been the subject of controversy in the debate over AI policy. Proponents of both approaches reference the constitutional clauses that support their claim. The reality is that this issue, as with many things in AI, is not a neat list of discrete policy matters. Instead, there is a clear mix of AI policy issues that require national attention—those that are the exclusive domain of the states and the issues that can be taken on by both the federal government and the states (or a collection of states). So how might a court or policymaker identify what level of governance is appropriate for different AI issues?

My paper, “The Obstruction Doctrine,” connects the jurisprudential dots that jointly establish a framework for evaluating whether a state’s exercise of power aligns with the letter and spirit of the Constitution. The first prong prevents states from inhibiting access to national markets. The second prong forecloses states from advancing laws that hinder the federal government’s ability to respond to issues that require national attention. And the third prong blocks states from interfering with national initiatives, such as extensive infrastructure projects. 

While no single source endorses this as a formal doctrine, a close inspection of overlooked federalism cases lends support to its various prongs. That said, this paper marks the beginning of a research inquiry into the contours and applicability of the doctrine rather than an announcement of a completed project. Additional inquiry will shine light on the descriptive accuracy of the doctrine—to what extent it aligns with how courts have reasoned through difficult cases, as well as its normative value, and to what extent its application aligns with and furthers the goals of our constitutional order. A review of the initial evidence behind the doctrine can guide that effort. 

The seed of the doctrine can be found most clearly in Buck v. Kuykendall. Writing for the Supreme Court, Justice Louis Brandeis struck down the law despite a long run of decisions upholding similar highway regulations—faulting it for dictating who could use the roads rather than how, for presuming to weigh the sufficiency of interstate competition, and for undercutting Congress’ own investment in federal aid highways. Each of these faults represents a prong of the doctrine.

Great A&P Tea Co., Inc. v. Cottrell is the first prong at work: The Supreme Court struck down a Mississippi law barring out-of-state milk unless the exporting state reciprocated, with Justice William Brennan seeing through the notion that a mutually agreed barrier is any less of a barrier. As he put it, such a scheme interferes with “the national interest in freedom for the national commerce.”

Under the second prong, a state may not seize a subject that requires one national rule. Arizona tried. It penalized railroads for running trains longer than 14 passenger or 70 freight cars and defended the cap as a safety measure Congress had never preempted. In Southern Pacific Co. v. Arizona, the Supreme Court held that train length was not Arizona’s to set: Where national uniformity is required, regulation must come from a single authority or not at all.

The third prong bars states from working against national efforts already underway. Iowa learned this in Kassel v. Consolidated Freightways Corp., in which the Supreme Court rejected a truck-length limit as a safety measure in name only and read the federal interest in open interstate transit off decades of investment rather than any one statute. Wisconsin learned it in Raymond Motor Transportation, Inc. v. Rice, when a rule so onerous that trucks detoured around the state fell for the same reason.

A handful of cases under a pithy title does not justify a recalibration of federalism jurisprudence. That is not my objective. This paper aspires to jump-start new conversations about how to resolve emerging, significant questions related to state and federal power. The Supreme Court has long discovered new frameworks as the nation’s economy, infrastructure, and technology shift. Wickard, Lopez, Raich, Pike, National Pork Producers, and other commerce clause cases, for instance, reflect the Court’s attempts to reconcile the Constitution’s principles with a world unanticipated by the Founders. Those cases all remain sound law. Yet, it would be hard to explain those opinions under a shared doctrine.

As the Supreme Court confronts a new set of federalism questions—from data centers to frontier AI governance—its habit has been to bend prior cases to fit contemporary politics and national priorities. This paper aims to get ahead of that tendency and explore how an approach can balance the lingering questions of federalism and prevent patchworks while still allowing policy experimentation when appropriate.