You have asked us to articulate the general legal principles reflected in case law under the so-called Dormant Commerce Clause, and to address briefly how those principles might bear on state laws seeking to regulate artificial intelligence (AI).

The Dormant Commerce Clause

The Constitution confers on Congress the power to “regulate Commerce . . . among the several States.” Art. I § 8, cl. 3. Though phrased as an affirmative grant of authority, the Clause “not only vests Congress with the power to regulate interstate trade; the Clause also contains a further, negative command, one effectively forbidding the enforcement of certain state economic regulations even when Congress has failed to legislate on the subject.” National Pork Producers Council v. Ross, 598 U.S. 356, 368 (2023) (cleaned up). That “negative command” is often referred to as the Dormant Commerce Clause, or DCC. Judicial articulation of the DCC’s contours has shifted over time, but case law generally recognizes it as encompassing three aspects:

• an anti-discrimination principle;

• an extraterritoriality principle; and

• an excessive-burden principle.

In Pork Producers, the Supreme Court recently described the anti-discrimination principle as lying “at the ‘very core’ of [its] dormant Commerce Clause jurisprudence.” Id. at 370 (quoting Camps Newfound/Owatonna, Inc. v. Town of Harrison, 520 U.S. 564, 581 (1997)). Broadly speaking, this principle prohibits States from enacting protectionist measures that benefit in-state interests at the expense of out-of-state competitors. Thus, state AI laws that are “driven by economic protectionism—that is, regulatory measures designed to benefit in-state economic interests by burdening out-of-state competitors,” are invalid under the DCC. Id. at 369 (Department of Revenue of Ky. v. Davis, 553 U.S. 328, 337–338 (2008)) (quotation marks and ellipses omitted).

But States, cognizant of this core principle, typically do not enact laws that are facially protectionist. As a result, state regulation of AI is more likely to be challenged under the DCC’s other two principles, which are explored at greater length below.

1. Extraterritoriality

The classic formulation of the prohibition on extraterritorial state regulation is that one State “has no power to project its legislation into [another State] by regulating the price to be paid in that [second] state for [products] acquired there.” Baldwin v. GAF Seelig, Inc., 294 U.S. 511, 521 (1935). That formulation has been taken more generally to mean that a State may not “regulate a commercial transaction that ‘takes place wholly outside of the State’s borders.’” Sam Francis Found. v. Christies, Inc., 784 F.3d 1320, 1323 (9th Cir. 2015) (en banc) (quoting Healy v. Beer Instit., 491 U.S. 324, 336 (1989)). For many years, this principle has been articulated as a categorical (per se) prohibition. Recently, however, the Supreme Court characterized its own anti-extraterritoriality case law as standing for a more limited rule—namely, that States may not enact “price control or price affirmation statutes that tie[] the price of in-state products to out-of-state prices.” Pork Producers, 598 U.S. at 374 (cleaned up).

At issue in Pork Producers was a challenge under the DCC to a California law that, among other things, banned the sale of pork products derived from breeding pigs in confined stalls. The plaintiffs, organizations whose members raise and process pigs outside of California, argued that the law was unlawful per se because of its “extraterritorial effects” on out-of-state pork production. Id. But the Supreme Court disagreed, noting that such extraterritorial effects were not unusual “[i]n our interconnected national marketplace,” where the laws of any particular State may “have a considerable influence on commerce outside their borders.” Id. at 374-75 (quotation marks omitted). Since the California law was neither a price control nor a price-affirmation statute, the Court explained, invalidating the law under the DCC merely because of the size and influence of the California market “would cast a shadow over laws long understood to represent valid exercises of the States’ constitutionally reserved powers.” Id. at 375.

In so ruling, however, the Court acknowledged that there may still be questions under the Constitution “about where one State’s authority ends and another’s begins—both inside and outside the commercial context.” Id. at 375. The Court indicated that a State’s attempt to regulate wholly out-of-state conduct might conflict with “the Constitution’s structure and the principles of sovereignty and comity it embraces,” particularly where the regulated conduct is “not intended to produce or . . . do[es] not produce detrimental effects within” the regulating state. Id. at 376 (cleaned up). In a notable footnote, the Court also indicated that even if such disputes do not “pose[] a dormant Commerce Clause question,” they might still “test[] the territorial limits of state authority under the Constitution’s horizontal separation of powers.” Id. at 377 n.1.

Following Pork Producers, while it remains unclear what sort of extraterritorial regulation is permissible, some general guidelines exist. On one end of the spectrum, a law’s practical out-of-state effects are not sufficient, by themselves, to invalidate a statute per se on extraterritoriality grounds. On the other end of the spectrum, price-affirmation statutes (i.e., laws that tie in-state prices in some fashion to the price of goods or services outside the State) are still unlawful. At least one court has also held, following Pork Producers, that any attempt to regulate the price of out-of-state sales—even if not through a price-tying regulation—is similarly unlawful. In Association for Accessible Medicines v. Ellison, 140 F.4th 957 (8th Cir. 2025), the Eighth Circuit invalidated Minnesota’s attempt to regulate the price of drug manufacturers’ sales of their products to wholesalers on the ground that those sales take place outside Minnesota. In so ruling, the court rejected the State’s argument that Pork Producers had “overturn[ed] [prior] cases that applied the dormant Commerce Clause to invalidate statutes that have the specific impermissible extraterritorial effect of controlling prices,” and also stated that “discrimination  is  not required  when a statute has the specific extraterritorial effect of controlling the price of wholly out-of-state transactions.” Id. at 961.

The status of laws in between these two extremes is less clear. Where a state law directly regulates out-of-state conduct—including where the law regulates conduct generally, without distinguishing between in-state and out-of-state conduct—it still may be unconstitutional as applied out-of-state, either under the DCC or under structural constitutional principles. That is especially so if the regulated conduct is not intended to have and does not have any significant in-state effects. But it remains unclear whether a State may directly regulate out-of-state conduct that does have in-state effects to some degree and, if so, what marks the relevant boundaries.

Insofar as AI regulation is concerned, state laws will be more or less vulnerable on extraterritoriality grounds depending on the in-state “hook” on which they are based. Where the state law directly regulates in-state conduct, the per se prohibition on extraterritorial state regulation will simply not apply (absent special circumstances, such as price-tying), even if the law has the practical result of affecting some out-of-state behavior. That means States are on relatively firm ground when they regulate AI as it has been developed, deployed, or sold within the State.

By contrast, where state law regulates out-of-state conduct, such as the out-of-state development, testing, or deployment of AI, the State will likely need to point to some in-state link. For example, imagine that a North Dakota law would regulate the development of AI applications—not merely the deployment of applications within the State—on the ground that North Dakota has an interest in the development of “safe” AI. In that circumstance, the law could be vulnerable under extraterritoriality (or analogous federalism) principles, since it regulates out-of-state conduct.

Pork Producers indicates that the extent of relevant in-state consequences may determine whether a state law’s extraterritorial reach renders it unconstitutional. Where the law regulates in an area in which there are significant in-state effects from AI that has been created or used out-of-state, the law may have a sufficient in-state connection to avoid concerns about extraterritoriality. But in situations where that out-of-state work has only minimal in-state effects, the fact that a company does some business within a State, has employees there, or has other nominal in-state connections may not be enough to permit regulation of the company’s out-of-state AI work. Cf. Sam Francis Found., 784 F.3d at 1320 (invalidating, on DCC grounds, California’s resale royalty statute, which required the seller of artwork originally produced by a Californian to pay a royalty to the author when the work was resold, even if resold in another State). In these “middle-category” cases, it is still too soon to tell how courts will approach the extraterritoriality principle after Pork Producers, and also whether they will pick up on the Court’s suggestion that concerns about state sovereignty might be better addressed under other constitutional principles (such as the separation of powers).

2. Excessive Burden

A separate line of cases applies to state statutes that regulate in-state conduct, but do so in a manner that imposes a burden on out-of-state commerce that is “clearly excessive in relation to the putatively local benefits.” Pike v. Bruce Church, Inc., 397 U.S. 137 (1970). In Pike, the Supreme Court invalidated an Arizona law requiring cantaloupes grown within the State to be packed and labeled in a certain manner approved by a state official, even if they were to be sold in other States. The Court struck down the law on the ground that Arizona had only a “tenuous interest” in having Arizona-grown cantaloupes packaged and labeled a certain way, compared to the very substantial burden that the law would impose on goods sold out-of-state. Id. at 145-46.

This test is known as “Pike balancing” or the Pike test: In evaluating state laws, “even nondiscriminatory burdens on commerce may be struck down on a showing that those burdens clearly outweigh the benefits of a state or local practice.” Dep’t of Revenue of Ky. v. Davis, 553 U.S. 328, 353 (2008). The classic Pike situations involve “state regulations on instrumentalities of interstate transportation,” such as state laws requiring interstate vehicles (like trucks or trains) to adopt certain safety features or other specifications when they travel through the enacting State. Pork Producers, 598 U.S. at 379 n.2; see, e.g., Bibb v. Navajo Freight Lines, Inc., 359 U.S. 520, 523-530 (1959) (state law specified certain mudflaps for trucks and trailers).

In Pork Producers, the Court fractured on the plaintiffs’ Pike challenge, with a majority of Justices holding that the California law at issue there did not fail under the Pike test, but under differing rationales. Writing for himself and two others (Justices Thomas and Barrett), Justice Gorsuch would have jettisoned the Pike test altogether, including because it requires comparing categories of benefits and burdens that may be incommensurate (e.g., the moral and health benefits of humanely treating animals versus the financial costs of regulation). Pork Producers, 598 U.S. at 380-83. But the other six Justices declined to join that portion of his opinion. As a result, Pike remains good law.

In another section of his opinion—this time writing for the Court—Justice Gorsuch concluded that California’s law “did not impose a sufficient burden on interstate commerce to warrant further scrutiny” under Pike. Id. at 384. Most notably, the law gave farmers a choice of how to comply with it: “They may provide all their pigs the space the law requires; they may segregate their operations to ensure pork products entering California meet its standards; or they may withdraw from that State’s market.” Id. That choice may well impose “some costs” of compliance on some producers, he explained, but many producers “will be able to pass along at least some of their increased costs to consumers,” and there was no plausible allegation that out-of-state consumers “will have to pick up the tab.” Id. at 386. The bottom line was that “harm to some producers’ favored ‘methods of operation’” is insufficient in itself to render a state law under Pike. Id. at 386-87 (quotation marks omitted). Writing for herself and Justice Kagan, Justice Sotomayor agreed with this portion of Justice Gorsuch’s opinion, though they did not support his view that Pike balancing is not a valid basis for evaluating a law’s constitutionality. Id. at 391-393.

Chief Justice Roberts, writing for himself and three others (Justices Alito, Kavanaugh, and Jackson), dissented. Id. at 394-403. This group of four Justices would have held that the plaintiffs had plausibly alleged a claim under Pike. In their view, the plaintiffs had “identif[ied] broader, market-wide consequences of compliance” with the California law, and these allegations were sufficient to make out a valid DCC claim under Pike. Id. at 397. Among other things, the plaintiffs alleged that compliance would increase production costs by more than $13 per pig—a 9.2% increase—requiring producers in aggregate to spend between $290 and $348 million. Id. at 399. In his view, the law also imposed other types of nonfinancial “harms,” such as requiring producers to adopt group housing, which could actually lead to worse health outcomes for the pigs, and potentially to the spread of diseases as well. Id. at 400. He also expressed concern that the law would “force compliance on farmers who do not wish to sell into the California market.” Id. at 401.

What is left of Pike balancing after Pork Producers? Because six Justices endorsed the Pike test, plaintiffs can still use Pike as the basis for a DCC challenge to the constitutionality of a state law. But the ability to invoke Pike does not guarantee success. Winning under Pike was often difficult even before the decision in Pork Producers. The test is designed to be deferential to States, and courts only rarely accepted such challenges—though they would often allow the plaintiff to go through discovery. The Pork Producers majority made winning such claims harder still, by seemingly rejecting as a categorical matter the significance of the plaintiffs’ attempted reliance on the costs of complying with the law. Because farmers could choose how to comply with the law—including by pursuing an option to segregate their production along state lines or exit from the California market altogether—the fact that the law made production substantially more expensive was insufficient to state a Pike claim, at least where the producers could pass on some of their increased costs to consumers. The Court also relied on the fact that some out-of-state pork producers actually welcomed the California law, to illustrate that the law merely disfavored certain production methods, as opposed to disfavoring out-of-state production generally.

Going forward, plaintiffs raising Pike claims will have to point to something more than mere compliance costs in order to invalidate a law. Based on Pork Producers, the best candidates are:

• laws that cannot be avoided by exiting the state market;

• laws that cannot be avoided by segregating production into compliant and noncompliant workstreams, such as where a single and uniform production method is required;

• laws that impose a burden that the regulated entity cannot pass on to its customers; and

• laws that impose burdens that fall disproportionately on out-of-state consumers (i.e., where the burden imposed on a particular out-of-state consumer is substantially greater than the burden imposed on an in-state consumer).

There is also an open question whether the Supreme Court would take account of non-economic burdens in Pike balancing. (Burdens that would independently rise to the level of a constitutional violation—for instance under the First Amendment—would require a separate analysis.). Chief Justice Roberts’s opinion indicates that at least four Justices are open to consideration of these burdens; the other Justices did not expressly address the issue.

Applying these principles to current AI proposals will require a case-specific analysis. The fact that a state law is triggered by in-state conduct does not immunize the law. For instance, in Pork Producers, the in-state trigger for California’s law—namely, a sale of meat within the State—did not automatically render the law permissible under Pike balancing. As a general matter, state laws regulating AI will be less vulnerable under the Pike test if their in-state effects are substantial in relation to any out-of-state effects, such as when they expressly apply only to conduct within the enacting State.

If a state law applied to out-of-state development, deployment, or sale of AI tools, however, then it would be more vulnerable. That is particularly so where there is no feasible way for the company to avoid the regulated state market. For instance, imagine that a State sought to regulate AI tools whenever they were “used” in the State—even if the developer never directly sold the tool or made it available to the State’s consumers—so long as the tool eventually made its way into the hands of state consumers, even if it got there solely as a consequence of consumer activity over which the developer had no control. In such a scenario, the company may have no way to segregate its production or prevent its tools from ending up in the regulating State. (The company in that case might also have an extraterritoriality claim akin to the one raised in the Association for Accessible Medicines, cited above, although the viability of such a claim in the case of non-price regulation is uncertain). By contrast, in circumstances where market segregation of products is possible—such as through geofencing—a Pike claim is relatively unlikely to succeed.

Finally, it is unclear how the Pike test would apply to state regulation of AI tools that are made available for free, in situations where an out-of-state developer cannot comply with the law without destroying the viability of the product. Open-source AI products might be one example. Pork Producers indicates that such a law might survive a Pike challenge on the ground that the developer’s choice not to charge for its product is merely a “particular . . . method[] of operation,” which the DCC “does not protect.” 598 U.S. at 384 (cleaned up). On the other hand, a developer challenging the law might have more success if it could show that there is no feasible way to “pass along” the law’s “increased costs to consumers.” Id. at 386 (cleaned up). The developer might also need to show that switching to an ad-supported or proprietary model is not feasible. As noted, successfully pursuing any Pike claim will depend on the specifics of the law at issue, the challenged application of that law, and the record created by the plaintiff to substantiate the law’s disproportionate out-of-state burden.

* * *

State laws seeking to regulate AI, like the technology itself, are still relatively new. The Dormant Commerce Clause provides some guidance on the limits of state AI regulation, but the contours of its current boundaries remain in significant flux.