The Trump Administration recently contemplated issuing an executive order aimed at combatting state restrictions on the development of artificial intelligence (“AI”), but has, for the moment ceased work on the executive order.  The Order, which could be resurrected, would have directed the Attorney General to assemble a task force of Department of Justice lawyers to challenge state laws regulating artificial intelligence, including mounting challenges based on the dormant commerce clause doctrine.  The Attorney General could pursue such an initiative even in the absence of an executive order. This three-part series suggests that even though private parties regularly bring dormant commerce clause claims challenging state laws that do not discriminate against interstate commerce, courts should hold that the Executive Branch may not do so. 

Part I of this series described the contemplated executive order and summarized dormant commerce clause doctrine.  Part II explained why courts, by entertaining some types of dormant clause suits brought by the Executive Branch, would undermine Congress’ preeminent role in regulating interstate commerce.  It ended by posing this question: what is the doctrinal hook for precluding the Executive Branch from bringing dormant commerce clause suits of a type typically brought by private parties?    This final installment of the series answers that question and discusses a few ancillary issues.

Dismissing Department of Justice Lawsuits Raising Dormant Commerce Clause Limitations on Political Question Grounds?

Congress, rather than the courts, should determine whether state AI regulation is imposing excessive burdens on interstate commerce.  One means of ensuring that occurs, judicial invocation of the non-justiciable political question doctrine to dismiss certain types of dormant commerce clause challenges brought by the Department of Justice, might appear quite odd.  See, Baker v. Carr, 369 U.S. 186 (1962); 13C FED. PRAC. & PROC. JURIS. § 3534.3 (3d ed.)(“Wright & Miller”)(available on westlaw)(exploring the political question doctrine).  Ordinarily the doctrine precludes any plaintiff from demanding that a federal court resolve an issue the plaintiff seeks to litigate.  See id.[1]  The standing, or perhaps the ripeness and mootness, doctrines are usually the justiciability doctrines deployed to prohibit only particular types of plaintiffs from raising an issue in Court.[2]  In this sense, invoking the political question doctrine to prevent the United States, but not private entities, from raising dormant commerce clause claims against non-discriminatory state regulations might seem inappropriate.[3] 

But the “political question doctrine” otherwise perfectly suits the approach I have outlined in the prior installments of this series, as well as the separation of powers rationale underlying that approach.  My suggested approach removes the courts as an avenue for the President to address non-discriminatory state regulation of interstate commerce when such regulation is not preempted by either federal legislation or a regulation authorized by congressional statute.  It thus ensures that issues involving non-discriminatory state regulation of interstate commerce remain subject to the give and take of legislative processes, preserving Congress’ preeminence in regulating interstate commerce. 

The political question doctrine has been viewed as one addressing separation of powers concerns, 13C FED. PRAC. & PROC. JURIS., supra, at § 3534.3,[4] even though sometimes the issues involved might masquerade as ones involving “the federal judiciary’s relationship to the States.”  See, Baker v. Carr, supra, 369 U.S. at 210.[5]   Thus, in Baker v. Carr, the Court observed: “The nonjusticiability of a political question is primarily a function of the separation of powers.”  369 U. S. at 210. It explained: “[i]n determining whether a question falls within [the political question] category, the appropriateness . . . of attributing finality to the action of the political departments and also the lack of satisfactory criteria for a judicial determination are dominant considerations.”  Id.

The President’s power to seek a judicial declaration that a state statute exceeds the constitutional bounds of state power to regulate the interstate commerce may appear to present an issue of the federal courts’ powers vis-à-vis state governments, or, in some respects, an issue of individual rights against state governments.[6] But, instead, it is fundamentally a question about the President’s, and the federal judiciary’s, power to cut Congress out of the regulation of interstate commerce. Even if such litigation does not threaten to completely exclude Congress from the business of regulating commerce, it would certainly, at the least, shift the considerable burden of overcoming legislative inertia from the President to Congress.[7]

Moreover, the major metrics used for determining the applicability of the political question doctrine, a textual commitment of an issue to another branch of government and the lack of a judicially definable standard, suggest that Department of Justice challenges to non-discriminatory state legislation on dormant commerce clause grounds present a non-justiciable political question.[8]  The issue of whether state regulation of aspects of AI programs excessively interferes with national policy toward the AI sector is textually committed to Congress, by the Interstate Commerce Clause.[9]  There is also, arguably, no judicially manageable standard for determining what the regulatory policy should be toward the nascent AI sector.  And the Court itself has often questioned whether the Pike test is one the judiciary has the capacity to perform.[10]

Brief Musings on Two Related Issues

Perhaps the Executive Branch could interpret some existing federal statutes as mandating or authorizing preemption of certain state statutes that would constrain the development of artificial intelligence systems.  For example, the draft executive order on AI suggested that the Chair of the Federal Trade Commission issue a policy statement explaining the circumstances in which state laws might be preempted by the Federal Trade Commision Act’s prohibition on engaging in deceptive acts or practices affecting commerce. See 45 U.S.C. §45(a)(1), Draft Executive Order, Eliminating State Law Obstruction of National AI policy, §6.[11]  What are the limitations on such an approach?

A complete discussion of the topic may well require some discussion of federal agencies’ authorization to preempt state laws and regulation on their own authority.[12] This post leaves that discussion for another day.  However, insofar as a federal agency purports to rely solely on a federal statute, unadorned by any regulation, as the basis of asserting preemption, it will face several constraints. 

First, there is, at least nominally, a presumption against finding obstacle preemption.  Rice v. Santa Fe Elevator Corp., 331 U.S. 218, 231 (1947); see generally, Gregory M. Dickinson, Calibrating Chevron For Preemption,  63 ADMIN. L. REV. 667, 673 (2011).[13] Granted, the presumption is often honored in the breach.[14]  Second, the Court has now abandoned the Chevron doctrine, reducing the level of deference courts will grant to agencies’ construction of the statutes they administer.[15]

Third, the “major questions doctrine” might well be applied in this scenario, or, alternatively the Court may craft a plain statement canon to constrain agencies’ abilities to interpret statutes as preempting non-discriminatory state regulation of commerce.[16]  Whether a statute, such as the Federal Trade Commission Act, broadly invalidates state regulation of AI is surely at least arguably a question of such great economic or social significance that the issue should be expressly addressed by Congress.  That is true all the more given Congress’ recent rejection of a moratorium upon state regulation of AI.  The potential existential threat of uncontrolled AI development further supports the proposition that AI is a major question with respect to any general federal statute, like the Federal Trade Commission Act, or any other statute not focused on AI.[17]

And, consistent with my argument above, a plain statement (or “clear statement”) canon might be appropriate in this context.   Clear statement rules provide that courts will not interpret a statute to contravene certain principles unless the legislature clearly indicates its intention to do so.[18]  For example, in Landgraf v. USI Film Products, 511 U.S. 244 (1994), the Court established a clear statement rule that statutes would not be interpreted to apply retroactively absent an express statutory statement of congressional intent to the contrary.  As Justice Kennedy explained in writing for the Court in Landgraf: 

Requiring clear intent assures that Congress itself has affirmatively considered the potential unfairness of retroactive application and determined that it is an acceptable price to pay for the countervailing benefits. Such a requirement allocates to Congress responsibility for fundamental policy judgments concerning the proper temporal reach of statutes. 

Id. at 272-73.[19]

Arguably, if Congress wishes a federal statute to displace state law based on the subtle type of balancing inherent in the Pike test, or even upon an assessment of whether the welter of state laws creates an unworkable patchwork for interstate commerce, Congress itself should explicitly say so.  Congress should perform the necessary balancing of interests or assessment of the degree to which the patchwork of state regulation is untenable.  Or, instead, Congres could explicitly and properly delegate that decision to an executive branch official.  A private party should be able to seek such a determination from the courts nevertheless, but the Executive Branch should not be free to pursue that approach rather than seeking congressional action, and acceding to the results.

Even if the Court were to adopt the approach that Executive Branch challenges to non-discriminatory state regulation of commerce are non-justiciable political questions, could Congress enact legislation reversing that judgment?  In other words, could Congress and expressly confer upon the Executive Branch the authority to invoke any available dormant commerce clause theory to challenge and obtain injunctive relief with respect to state regulation of artificial intelligence? The simple answer might appear to be yes.  After all, federal courts already entertain such challenges by private entities.  And it would seem that Congress could, if it so chose, deputize the Executive Branch to protect the national markets from unduly disruptive state regulation which threatens to greatly inhibit, or destroy altogether, some segment of national, and indeed international, commercial activity.

But is there a sort of non-delegation doctrine problem?  Again, at first blush, no.  The non-delegation doctrine is focused upon delegating the power to legislate to the executive branch, not the authority to exercise the executive powers of the United States.  But such a move would seem to allow the President to declare and enforce a national policy vis-à-vis states that Congress has never adopted.  This might perhaps give rise to a sort of non-delegation doctrine prohibiting the President from, in effect, making law by selective litigation against state laws regulating artificial intelligence.

Conclusion

The breathtakingly rapid development and deployment of artificial intelligence products and technology would seem to demand some formulation of national policy, and quickly.  Yet the federal legislative process is often frustratingly slow and Congress often fails to keep pace with changes in technology.  However, under our form of government, it is Congress that is to make fundamental decisions about the regulation of artificial intelligence, and the role states may play in the absence of congressional action.  Entertaining suits brought by the Executive Branch challenging non-discriminatory state statutes on the basis of dormant commerce clause doctrine threatens to undermine that Congressional role.

---

[1] 13C FED. PRAC. & PROC. JURIS., supra, at § 3534.3 (“[t]he clearest distinction between political-question doctrine and other justiciability doctrines is the conclusion that a political question is never suitable for judicial disposition, no matter what individual injury is pressed.”)

[2] The standing doctrine seeks to identify particular plaintiffs who lack a sufficiently concrete injury, attributable to the defendant governmental entity’s allegedly unlawful action, for which a court could provide redress. Lujan v. Defenders of Wildlife, 504 U.S. 555, 560-61, 573-78 (1992); 13C FED. PRAC. & PROC. JURIS., supra, at §3534.3 (see text accompanying notes 24-25).  But one aspect of the doctrine, seen most dramatically when the doctrine appears to preclude anyone from challenging a defendant government entity’s decision, reflects separation of powers concerns.  Valley Forge Christian College v. Americans United for Separation of Church and State, 454 U.S. 464, 473 (1982); see, U.S. v. Richardson, 418 U.S. 166 (1974)(“the absence of any particular individual or class to litigate these claims gives support to the argument that the subject matter is committed to the surveillance of Congress, and ultimately to the political process”). 

The ripeness and mootness doctrines focus on questions of timing, but the timing questions that are often unique to the particular plaintiff.  See, 13C FED. PRAC. & PROC. JURIS., supra, at § 3534.3. Thus, a case may be deemed unripe because the particular plaintiff’s challenge is premature.  A case may be declared moot because the particular plaintiff’s challenge comes too late, and whatever relief might have been granted at an earlier time is now useless or inconsequential.

[3] Note that a presidential legislative initiative is far more likely to reach the legislative agenda for serious consideration, such as a vote in the House or Senate chamber or a cloture motion to break a filibuster, than is an initiative advocated solely by business entities.  Moreover, the commercial interests of businesses and business associations are far more parochial than the Executive Branch’s interest in safeguarding the nation’s regulatory authority, or the federal government’s ability to engage in regulatory forbearance.

[4] 13C FED. PRAC. & PROC. JURIS., supra, at § 3534.3 (noting that even though decisions in cases involving the political question doctrine are often “expressed in terms of limitations on judicial power,” it is obvious that “the limitations are intimately tied to the substantive powers of the other branches”).

[5] Baker v. Carr, supra, 369 U.S. at 210 (noting that its review of political question doctrine precedents “reveal[ed] that [in a large class of political question cases] it is the relationship between the judiciary and the coordinate branches of the Federal Government, and not the federal judiciary’s relationship to the States, which gives rise to the ‘political question’” (emphasis added)). 

[6] See, Bernard W. Bell, Eliminating State Law “Obstruction” of National Artificial Intelligence Policy ─ Part II, n.1, YALE J. REG.: NOTICE & COMMENT BLOG (Dec. 5, 2025).

[7] Id., n.3; see generally, id  (“[n]o private entity pursuing tis own commercial interests can command the deference courts accord a presidents’ assertions of his views of the national interests”).

[8] Baker v. Carr, supra, 369 U.S. at 210.  Later in its opinion, the Court sets forth a six-factor test.  Id. at 217.

[9] A skeptic might note the foreign commerce clause power, which Article I, section 8 grants to Congress in very similar terms as it grants Congress the power over interstate commerce.  Does that suggest that the President cannot initiate litigation to defend the national interest in terms of state interference with foreign commerce? The analysis might differ in terms of the power over foreign commerce, given the President’s greater role in foreign affairs, U.S. v. Curtiss-Wright Export Corp., 299 U.S. 304, 319-22, (1936), and Congress’ typically broader delegations in that area.  See, e.g., Crosby v. National Foreign Trade Council, 530 U.S. 363, 374-77 (2000)(invalidating Massachusetts law barring state entities from buying goods or services from companies doing business with Burma).  Moreover, in explicating the political question doctrine itself, the Court has recognized that questions of foreign policy are prominent in most cases in which the political question doctrine has been successfully invoked. Baker v. Carr, supra, 369 U.S. at 211.

[10] Eliminating State Law Obstruction ─ Part II, supra (see extensive discussion of the Justices’ respective positions in National Pork Producers Council v. Ross, 598 U.S. 356 (2023)).

[11] In particular, the draft Order suggests that regulation of algorithmic discrimination may be preempted, given that, in the President’s view, such statutes may require alteration of truthful outputs produced by AI models.

[12] See, Nina A. Mendelson, A Presumption Against Agency Preemption, 102 NW. U. L. REV. 695 (2008) (arguing for limiting agencies’ authority to preempt state law); Joshua Hawkes & Mark Seidenfeld, A Positive Defense of Administrative Preemption, 22 GEO. MASON L. REV. 63 (2014) (taking a contrary position); Catherine Sharkey, Inside Agency Preemption, 110 MICH. L. REV. 521 (2012)(providing case studies).

[13] The Court asserted: “we start with the assumption that the historic police powers of the States were not to be superseded by the Federal Act unless that was the clear and manifest purpose of Congress.”

[14] Id. at 681 (observing “[i]n any given preemption case it is nearly impossible to predict whether the [Santa Fe] presumption will make an appearance[;] [c]ommentators have long criticized the Court’s halfhearted and haphazard application of the doctrine”).

[15] However, even under the Chevron regime, the Court seemed disinclined to give Chevron deference to agency determinations that a statute the agency administered preempted state law.  See, Gregory M. Dickinson, Calibrating Chevron For Preemption, 63 ADMIN. L. REV. 667, 680-82 (2011).

[16] The major questions doctrine provides that “where the ‘history and the breadth of the authority’ an agency asserts is coupled with the ‘economic and political significance’ of the potential exercise of such authority, a court should ‘hesitate before concluding that Congress’ meant to confer such regulatory authority.”  West Virginia v. EPA, 597 U.S. 697, 721, 142 S. Ct. 2587, 2608 (2022); Resurrecting Skidmore in a New Era, supra, 55 SETON HALL L. REV. at 1505-1602.

[17] Does AI Pose An Existential Risk? We Asked 5 Experts, THE CONVERSATION  (Oct. 5, 2025); Nir Eisikovits & The Conversation US, Commentary: AI Is an Existential Threat—Just Not the Way You Think, SCIENTIFIC AMERICAN (July 12, 2023); Mark MacCarthy, Commentary: Are AI existential risks real—and what should we do about them? (Brookings July 11, 2025).  For a broader though somewhat dated view of the promise and problems of algorithms and artificial intelligence, see Patrick Marshall, Algorithms and Artificial Intelligence: Are they being used in harmful ways? (CQ Press July 6, 2018)(behind paywall).

[18] Bernard W. Bell, Using Statutory Interpretation to Improve the Legislative Process: Can it be Done in the Post-Chevron Era?, 13 J.L. & POL. 105, 136 (1997); William N. Eskridge, Jr. & Philip P. Frickey, Quasi-Constitutional Law: Clear Statement Rules as Constitutional Lawmaking, 45 VAND. L. REV. 593, 629 (1992)  (observing, in the early 1990’s, that “the current Court is unusually activist in the way it does statutory interpretation, creating clear statement rules and super-clear statement rules as means by which the Court can read constitutional values into statutes, albeit with the possibility of congressional override”).

[19] Perhaps the Santa Fe presumption was envisioned as a “clear statement” doctrine before clear statement doctrines became “cool.”  (The reference, as some of you surely know, is to Barbara Mandell’s popularization of a country music song entitled “I Was Country When Country Wasn’t Cool” in 1981.)  But the consistency with which the Santa Fe presumption has been honored in the breach suggests that if it was intended to require an express congressional statement of intent, it does not function that way now.