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 its efforts.  The order would have directed that Attorney General to assemble a task force to challenge state laws regulating artificial intelligence, including mounting challenges based on the dormant commerce clause doctrine.  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 provided an overview of dormant commerce clause doctrine.  It ended by posing this question: why should the United States be treated differently than the private entities that routinely mount dormant commerce clause challenges against state legislation?  This second installment of the series answers that question.

Why Shouldn’t the Attorney General Be Able to Bring At Least Some Dormant Commerce Clause Challenges in the Name of the United States?

Dormant commerce clause cases brought by private entities lack the implications for the separation of powers between the President and Congress implicit in such suits brought by the Department of Justice to vindicate “federal” interests.  Some dormant commerce clause suits by the executive branch should be permissible; others should not.

Claims seeking to address state discrimination against interstate commerce should be addressable by the Department of Justice.  The Supreme Court has quite properly crafted a default rule that, in the absence of congressional action, states may not discriminate against out-of-state commerce or out-of-state market participants.[1]  Thus, Department of Justice litigation seeking to enjoin such regulations does not undermine Congress’ implicit embrace of the dormant commerce clause doctrine’s anti-discrimination principle.  And dormant commerce clause litigation invoking the anti-discrimination principle is particularly important in disciplining state protectionism.[2]

Department of Justice litigation seeking to invalidate state legislation under the Pike v. Bruce Church or the incompatible state regulation strands of dormant commerce clause doctrine has quite different implications.  As noted previously, the Constitution explicitly grants Congress the power to regulate interstate commerce by means of legislation.  The President does have a formal role in that legislative process, captured by the Presentment Clause, U.S. Const., art. I, §7, cl.2.  However, in seeking legislation the President must persuade Congress to take the initiative.  Indeed, Article I, section 7 creates a regime that makes Congress the driving force in the legislative process, with both the first, and the last, word on legislation.[3]

Both Pike v. Bruce Church and conflicting regulatory regime problems, which involve facially non-discriminatory state legislation, must ordinarily be resolved by Congress through the legislative process.  Courts rarely uphold dormant commerce clause challenges based on such grounds.  That is as it should be.  The questions that must be resolved under those strands of the dormant commerce clause doctrine frequently yield neither clear-cut nor easily definable answers.  Rather they involve the weighing of state interests furthered by the regulation against the federal interests in either a uniform national market or protecting industries from unnecessary state interference.  Congressional inertia, of course, is a major malady of our current political system.[4] But such inertia presents a somewhat lesser challenge with regard to potential congressional action that merely limits state interference with interstate commerce.  Congress need not develop or agree upon an affirmative program of regulation.[5] Congress need only either divest states of regulatory authority indefinitely or for a more limited time period (i.e., by way of a moratorium) while it contemplates it own approach.[6]

Resolution of such issues is so fraught with difficulty for courts that every currently serving Justice has acknowledged the difficulties inherent in applying the Pike v. Bruce Church test, and several have questioned its legitimacy altogether.  In Ross, three Justices asserted that the test was not a legitimate one for the Judiciary to apply, likening it to discredited Lochner Era judicial doctrine.  Ross, supra, 598 U.S. at 380.  To paraphrase Justice Gorsuch: “In a functioning democracy, policy choices,” involving incommensurate social and economic costs and benefits, like those required in performing Pike balancing, “usually belong to the people and their elected representatives.” Ross, supra, 598 U.S. at 382.  Justice Gorsuch’s attack on the Pike test’s legitimacy and his appeal to overrule it were too extreme for six of his colleagues.  But his argument that that judges should be skeptical of their capacity to weigh incommensurable values was more appealing to some of his colleagues.[7]

Chief Justice Roberts, writing for himself and Justices Alito, Kavanaugh, and Jackson, acknowledged the validity of the Gorsuch group’s concerns regarding the Judiciary’s capacity to perform the balancing of incommensurable values required by the Pike test, but concluded that “sometimes there is no avoiding the need to weigh seemingly incommensurable values.” Id. at  396 (Roberts, J. concurring).  Thus, the Roberts group acknowledged the difficulty judges face in making the judgments required by the Pike test, and seemed to regard doing so as a last resort. 

And Justice Sotomayor, for herself and Justice Kagan rejected the proposition that applying Pike to a truly non-discriminatory state law was never appropriate. However, they did acknowledge that the inquiry often called for by Pike balancing, namely “balancing economic burdens against noneconomic benefits,” is “difficult and delicate, and federal courts are well advised to approach the matter with caution.”  Id. at 392 (Sotomayor, J., concurring).  She embraced the alternative path to avoid engaging in Pike balancing offered by Justice Gorsuch’s opinion for the Court, namely declaring that plaintiffs had failed to “allege a substantial burden on interstate commerce,” and thus failed to surmount the threshold for requiring balancing of the burden on commerce against the putative local interest. Id. at 383-87 (Opinion of the Court); id. at 393 (Sotomayor, J., concurring). 

Concerns about the Pike test have long been expressed in judicial opinions and the scholarly literature.  See, The Internet and the Dormant Commerce Clause, 110 YALE L. J. at 820 (citing sources).

Moreover, challenges to state regulation of commercial activity by the United States will inevitably alter the pro-state presumptions expressly set forth in the Pike test.  No private entity pursuing its own commercial interests can command the deference courts accord a president’s assertions of his view of the national interest.  Suits brought by the Attorney General to vindicate the President’s view of the national interest will thus inevitably begin to tip the balance in such cases much more in favor of invalidating rather than upholding the state regulation.

Thus, non-discriminatory regulations that have interstate impacts should be addressed by Congress, through the legislative process, and not by courts, at least not at the behest of the President of the United States or his Attorney General. Such suits will generally not claim any loss, in terms of the federal government’s proprietary interests, as a result of state regulation.  Moreover, the legislative process is arguably the process more protective of the state’s federalism interests, which should be carefully considered in resolving conflicting authority over interstate commerce.[8] The legislative process is also far more transparent than the processes for promulgation of executive orders.[9]  And given the limitations of adjudication as a means of quickly crafting a comprehensive regulatory approach, legislative action is more likely to produce a comprehensive and nuanced approach to reconciling federal and state regulatory authority over AI than a campaign of litigation.

Two Suggestive Precedents: The Steel Seizure Case and the Pentagon Papers Case

Two landmark cases, Youngstown Sheet & Tube Co. v. Sawyer, 343 U.S. 579 (1952)(“the Steel Seizure Case”), and New York Times v. United States, 403 U.S. 713 (1971)(“the Pentagon Papers Case”), offer perspective on the separation of powers concerns discussed above.  The Steel Seizure case emphasizes the importance of the Senate’s rejection of a provision that would have imposed a ten-year moratorium on state regulation of AI, which closely resembles the President’s contemplated AI policy.  Separate opinions in the Pentagon Papers case suggest that courts should not entertain certain types of Executive Branch litigation unauthorized by statute.

The Steel Seizure Case

As most readers of this post probably know, the Steel Seizure case involved President Truman’s seizure of the nation’s steel mills during a labor dispute so as to maintain the production of steel needed to sustain the nation’s war effort in Korea.  Much like President Trump does now, President Truman framed the move as one essential to national security.  The Court held that the President’s war powers could not extend to asserting such control over domestic industries, even industries with national security implications.  Youngstown Sheet & Tube Co. v. Sawyer, supra, 343 U.S. at 587.

But in an influential separate concurrence, Justice Robert Jackson noted that the President’s implied powers reached their nadir when Congress had refused to confer such power on the President and instead established another regime to address labor strife in important national industries.  Id. at 637-39.  He explained that when the President acts in such circumstances, he must rely only on those powers conferred by the U.S. Constitution.  Id. at 637-38. 

Here, the policy President Trump contemplated announcing stands in direct contrast to congressional action refusing to adopt a moratorium on state regulation of artificial intelligence.  Indeed, the President seems to have taken the very proposal Congress rejected, and transformed it into an executive order.  Granted, Senate rejection of a proposal, even by a vote of  99-1, cannot carry the weight of bicameral adoption of an affirmative statute to address an issue (as was the case in the Steel Seizure Case). Nevertheless, it would seem strong indication of legislative sentiment against the policy the President considered adopting in the draft executive order.[10] 

And President Trump’s national security argument seems somewhat weaker than that of President Truman.  Winning a technological race is not so directly critical to national security as the ability to prosecute an active foreign military conflict.  The latter relates to the President’s role as Commander in Chief to prosecute a war, the latter has a more tenuous relationship to any power exclusively conferred upon the President by the Constitution.[11]  The inevitable accretion of power toward presidents based on a national security rationale in the post World War II era was a concern of the majority in the Steel Seizure case.[12] 

The Pentagon Papers Case

In the Pentagon Papers case, the Nixon Administration sought to enjoin the publication of classified documents that Daniel Ellsberg, a defense contractor, had leaked to the New York Times and the Washington Post.  The Court issued a per curium opinion announcing that Court’s conclusion that the Administration had failed to meet the demanding standards for imposing a prior restraint on the press.  New York Times v. United States, supra, 403 U.S. at 714.  But Justices Stewart, White, and Marshall elaborated upon the separation of powers implications inherent in the Government’s request that the federal judiciary invoke its equitable powers to enjoin publication of information damaging to national security.  Id. at 730 (Stewart. J. concurring); id. at 729-33, 740 (White, J., concurring); id. at 741-46  (Marshall, J., concurring).  Perhaps, Justice Marshall framed the separation of powers concerns most clearly.

He noted the Government’s assertion that the Government’s inherent power to protect itself and the President’s powers to conduct foreign affairs and command the military “give him authority to impose censorship on the press to protect his ability” to deal effectively conduct the foreign nations and military affairs of the country.  Id. at 741.  But this was not permissible for two reasons.

First, he noted, the Court’s use of its contempt powers to prevent behavior that Congress has specifically declined to prohibit would “be utterly inconsistent with the concept of separation of powers.”  Id. at 742.  And indeed, he found that to be the situation before the Court, observing that “on at least two occasions Congress has refused to enact legislation that would have made the conduct engaged in here unlawful and given the President the power that he seeks in this case.”  Id. at 746.

But Justice Marshall offered a second rationale as well.  Similar harm to constitutional checks and balances between the three co-equal branches of Government would occur “even where Congress has granted the Executive Branch “adequate authority” to protect national security, but the executive has “choose[n] instead to invoke the contempt power of a court to enjoin the threatened conduct.”   Id. at 742.  The Constitution did not provide for “government by injunction,” which would permit the courts and the Executive Branch “make law” without regard to Congress’ lawmaking powers. Id.  And while “[i]t may be more convenient for the Executive Branch to need only convince a judge to prohibit conduct rather than ask the Congress to pass a law,” convenience does not justify such “a basic departure from the principles of our system of government.”  Id.

Or, as Justice Byron R. White put it:

At least in the absence of legislation by Congress, based on its own investigations and findings, I am quite unable to agree that the inherent powers of the Executive and the courts reach so far as to authorize remedies having such sweeping potential for inhibiting publications by the press. . . . To sustain the Government in these cases would start the courts down a long and hazardous road that I am not willing to travel, at least without congressional guidance and direction.

Id. at 732 (White, J., concurring)(emphasis added).

Dormant commerce clause litigation differs from efforts to impose prior restraints on the press by judicial degree.  Prior restraints are the most constitutionally disfavored mechanism for restraining speech.  See, John Calvin Jeffries, Jr., Rethinking Prior Restraint, 92 YALE L.J. 410, 412-19 (1983).[13]  Yet the dormant commerce clause doctrine raises quite important issues that go to the heart of the constitutional balance between the three branches of government.  So here, as with respect to the imposition of prior restraints against the press, there are reasons for courts to insist that Congress authorize the Department of Justice to bring suits to invalidate certain state regulations of commerce on dormant commerce clause grounds.

A Remaining Question: Previewing Part III

But what is the doctrinal hook for precluding the Executive Branch from bringing dormant commerce clause suits of a type typically brought by private parties?  My perhaps surprising answer, and a defense of it, awaits in the final post of this series.

---

[1] The prime rationale underlying the dormant commerce clause doctrine generally, and the anti-discrimination principle in particular, was perhaps most eloquently stated by the first Justice Jackson in H.P. Hood & Sons v. DuMond, 336 U.S. 525, 539 (1949). Accord, National Pork Producers Council v. Ross, 598 U.S. 356, 395 (2023) (Roberts, J. concurring).  But a second rationale brings an individual rights perspective to the dormant commerce clause restrictions, at least with regard to statutes that discriminate against out of staters.  The anti-discrimination principle protects out of state interests that would not otherwise be protected in the State’s political processes, Southern Pacific Co. v. Arizona, 325 U.S. 761, 768 n.2 (1945); see also Ross, supra, 598 U.S. at 370, 384-85 (contrasting the case before it, in which there are affected in-state producers, with Exxon Corp. v. Governor of Maryland, 437 U.S. 117 (1978), a case in which there were no in-state producers that would be covered by Maryland’s statutory limitations);  cf. McCulloch v. Maryland, 17 U.S. 316, 428-429 (1819)(while the interests of state legislators and constituents influence over their representatives can be relied upon to ensure state legislatures do not impose abusive taxes on their citizens, the same cannot be said of the Maryland legislature’s imposition of a tax upon federal instrumentalities, such as the Second National Bank of the United States, an entity representing the interests of all people of the United States, including those who are not Maryland citizens); see generally, Brannon P. Denning, Reconstructing the Dormant Commerce Clause Doctrine, WM. & MARY L. REV. 417, 481-84 (2008).

[2] Southern Pacific, supra, 325 U.S. at 768 (citing cases)(“[w]hether or not this long-recognized distribution of power between the national and the state governments is predicated upon the implications of the commerce clause itself, or upon the presumed intention of Congress . . . the result is the same”).

[3] See, Bernard W. Bell, Dead Again: The Nondelegation Doctrine, the Rules/Standards Dilemma, and the Line Item Veto, 44 VILLANOVA L. REV. 189, 220-224 (1999)(discussing Congress’ “last word” power in the context of the line-item veto). Indeed, the President essentially proposes to “flip” the Article I, section 7 script.  Under the proposed approach, the President would set the policy, to be enforced by litigation and funding conditions explicitly rejected by at least one house of Congress.   States could regain the power to regulate the artificial intelligence sector only by persuading Congress to pass legislation explicitly allowing them to do so.  To do so Congress would either need to obtain the President’s concurrence or amass the supermajority needed to override the President’s veto.  So long as 40 senators are willing to filibuster such legislation and/or one-third of members of either the House or the Senate are willing to uphold a presidential veto, such legislation cannot be adopted.

[4] Bernard W. Bell, Loper Bright: Resurrecting Skidmore in a New Era, 55 SETON HALL L. REV. 1577, 1580, 1595 (2025).

[5] One reason for legislative inaction on a particular subject is the difficulty of forming a consensus on an approach to address a problem.  In other words, large majorities may agree that a problem needs to be addressed, but find themselves unable to agree upon a positive political program to address the issue.  See, Resurrecting Skidmore, supra,55 SETON HALL L. REV. at 1600-01.

[6] Even then, questions of the carve-outs for the ban or moratorium may lead to disagreements, deadlock, and, consequently, inaction.  See, Resurrective Skidmore, supra, 55 Seton Hall L. Rev. at 1600-01 (describing the problems that can lead to deadlock or inaction); Johnson v. Transportation Agency, Santa Clara Cnty, 480 U.S. 616, 671 (1987) (Scalia, J., dissenting)(same).

[7] Justice Barrett wrote separately to emphasize the incommensurability of the values plaintiffs were asking the Judiciary to weigh.  Ross, supra, 598 U.S. at 393-94 (Barrett, J. concurring).

It is interesting to speculate about how Justice Gorsuch came to author the opinion for the Court.  Given the final line-up of the Justices, the Court’s post-argument conference may well have revealed a 5-4 schism regarding the appropriateness of applying the Pike balancing test to the California statute, with the Chief Justice being in the minority.  That would mean that Justice Thomas, the Court’s most senior Associate Justice, would get to assign the case.  But Justice Thomas believes that the dormant commerce clause doctrine itself is illegitimate.  It would seem sensible for him to assign the case to some other Justice, and Justice Gorsuch’s and Barrett’s views were presumably more in line with his own than were those of Justices Sotomayor and Kagan.

[8] The substantive law, namely dormant commerce clause doctrine, is quite protective of state interests, and thus, in theory, any Department of Justice lawsuit seeking to invalidate a state regulation would need to surmount a high hurdle.  But, as noted above, there are reasons to believe that the weighing of interests under the Pike test will tip decidedly against states in suits brought by the Attorney General to vindicate the national interest.

[9] While the process of proposing, working up, amending, and debating legislation is quite open, the process for the President establishing executive orders is very much the opposite, Congressional Research Service, Executive Orders: An Introduction 2-4 (March 29, 2021).  Indeed, the draft AI executive order seeking to curtail state authority came to light only due to a leak.

[10] The potential for presidents to use executive orders when they cannot persuade Congress to enact their legislative agenda is manifest.  See, Jeffrey A. Fine & Adam L. Warber, Circumventing Adversity: Executive Orders and Divided Government, 42 PRESIDENTIAL STUDIES QUARTERLY 256, 258-61, 272 (2012). 

Executive Order No. 12,954, which President Clinton promulgated in 1996, and directed federal agencies to refrain from contracting with companies that “permanently replace lawfully striking employees,” provides one example. The directive, nominally directed at federal agencies, was intended to establish a “balance” between worker and employers in the private sector. Chamber of Commerce v. Reich, 74 F.3d 1322, 1332–33 (D.C. Cir. 1996).  Indeed, Congress had considered previously considered enacting legislation under its Commerce Clause powers to accomplish the same goal by similar means, prohibiting companies from permanently replacing striking employees.  Id. at 1325. The efforts included the Workplace Fairness Act introduced in the U.S. Senate in President Clinton’s first full day in office.  Id.;  S.55, 103rd Cong., 1^st Sess.  (unenacted)(introduced on January 21, 1993).  The bill was stymied by the filibuster, when two cloture motions failed to succeed in breaking the filibuster.  Record Votes 188 & 189.    The Executive Order was ultimately struck down by the D.C. Circuit as conflicting with the provisions of the National Labor Relations Act. Chamber of Commerce v. Reich, supra, 74 F.3d at 1332-39.

[11] This is in no way to suggest that the national interest in prevailing in the competition regarding artificial intelligence is not extremely important, or that quite dire consequences might flow from losing that race.

[12] See, Youngstown Sheet & Tube Co. v. Sawyer, supra, 343 U.S. at 587 (acknowledging that the “theater of war” is an expanding concept).  Certainly, the dissent emphasized the Cold War atmosphere and the nation’s global responsibilities.  Id. at 669-70 (Burton, J., dissenting).

[13] Rethinking Prior Restraint, 92 YALE L.J. at 410 (observing that, properly conceived, the “prior restraint” doctrine “imposes a special disability on official attempts to suppress speech in advance of publication ─ a disability that is independent of the scope of constitutional protection against punishment subsequent to publication,” such that “speech that validly could be controlled by subsequent punishment nevertheless would be immune from regulation by prior restraint”).