The Trump administration’s announcement in May of a nearly $1.8 billion “Anti-Weaponization Fund”—created (though later blocked and perhaps abandoned) as part of a settlement of President Trump’s lawsuit against the IRS—raised questions about whether the parties in the lawsuit were genuinely adverse, whether the court had jurisdiction, and what role courts have in policing potentially collusive settlements. Underlying these questions is a more fundamental one: what led the Department of Justice to pursue and approve this settlement?

As striking as the Weaponization Fund is, it is one example in a broader and growing pattern. Across a range of subject areas, the second Trump administration has pushed past the norms that have historically governed how the Department of Justice approaches settlements and changes in litigation position.

In a new paper, I have identified five such norms based or reflected in the Justice Manual, longstanding DOJ guidance, Supreme Court precedent, and institutional practice developed across administrations of both parties. I also discuss several examples and preview how this pattern may continue, particularly in the environmental sphere.

The Norms and Their Sources

The most foundational norm is that settlements are based on the public interest, not politics. The Justice Manual, a guidance document for Department of Justice attorneys, makes apolitical legal judgment a “fundamental duty of every employee” and lists factors such as litigation risk, cost, and the seriousness of the violation as appropriate settlement considerations. DOJ processes help reinforce this norm: for example, the centering of recommendations developed by career attorneys in the settlement process.

A second norm is that litigation and settlement are conducted between genuinely adverse parties, and that settlements do not promise substantive policy results that bypass required administrative processes. A 1986 memo from Attorney General Meese and a 1999 Office of Legal Counsel opinion by Acting Assistant Attorney General Randolph Moss both conclude that DOJ should ordinarily not enter settlements that commit agencies to such outcomes. Third, and closely related, is the norm extending from the Meese and Moss memos that DOJ does not enter settlement agreements that constrain successor administrations’ discretion.

A fourth norm is that settlements fit the dispute: they should not extract promises unrelated to the case at hand, and they should not be broader than the claims the court has jurisdiction to resolve.

A fifth governs changes in litigation position. The bar for changing position mid-litigation is high, and the process matters: changes must be well-reasoned, the Solicitor General’s office follows a deliberate multi-layer process before confessing error, and changes in position should not be used to achieve through courts what should happen through administrative processes.

A Pattern of Departures

Recent examples show departures from these norms, often in service of deregulatory outcomes, promoting preferred industries or interests, and making rapid policy change. A few examples:

• In a recent immigration case, the federal government and the State of Florida—aligned parties after the change in administration—reached a judicially enforceable agreement barring future administrations from exercising statutory parole authority in a particular way until 2041. That consent decree runs counter to the longstanding policy that one administration does not make policy commitments that will bind the next.
• EPA confessed error in a case about a Biden-era air pollution standard (for particulate matter), asking the D.C. Circuit not merely to remand the rule to the agency for reconsideration but to vacate it outright—achieving through judicial order a deregulatory change that would otherwise require a notice-and-comment rulemaking in which affected parties could participate. (Just recently, the D.C. Circuit declined that request, rejecting EPA’s new post change-in-administration position that the rule was unlawful).
• The administration reached pre-litigation “settlements” with offshore wind developers, requiring them to surrender their leases in exchange for federal payments; the companies must make equal investments into oil and gas projects. In addition to politicization concerns, this example raises questions about whether the settlements fit the allegedly threatened litigation (and whether the threat of litigation was genuine). A coalition of states have now sued to block one of the agreements.

American Gas Association v. DOE: A New Example?

Given this pattern, a recent DOJ filing at the Supreme Court warrants examination.

In 2022, industry groups challenged a Biden-era Department of Energy rule tightening efficiency standards for certain natural gas furnaces and water heaters (that rule reversed a Trump-era rule; at the heart of the change was whether certain technology constituted a “performance characteristic” that DOE could not regulate out of existence under the Energy Policy and Conservation Act). DOJ, representing DOE, defended that rule through briefing and oral argument in November 2024.

The D.C. Circuit upheld the rule in a 2-1 decision in November 2025, holding that the EPCA left DOE a “degree of discretion” under Loper Bright and that DOE had reasonably concluded that certain features of regulated equipment were not a “performance characteristic” under the statute. The dissenting judge accused the majority of doing an end run around Loper Bright and improperly deferring to the agency.

Industry groups filed a petition for certiorari in January 2026. But instead of filing a response urging the Court to deny review (or waiving its response, as the government often does), DOE, through DOJ, filed a brief asking the Supreme Court to grant, vacate, and remand the case (“GVR”), arguing that the D.C. Circuit had erred in its statutory interpretation by adopting an interpretation the government itself had advanced. The Supreme Court granted that request without explanation on June 8, 2026.

In its brief, DOE stated that its own December 2021 rule rested on a legal error as to the meaning and scope of the “performance characteristics” limitation in EPCA. It urged that “[b]ecause the government agrees that the rules at issue rest on legal error, the appropriate resolution is to GVR,” citing the Court’s “broad power” to vacate and remand cases, including where the Solicitor General has confessed error or an agency has a new interpretation of a statute. DOE had never signaled to the D.C. Circuit between the change in administration and the court’s decision 10 months later that it had changed its mind about the rule, nor had DOE indicated an intent to revise or rescind the rule through an administrative process. Nevertheless, DOE’s brief supporting GVR stated that “the agency is considering a new rulemaking in which it would correct” factual and legal errors, and it suggested that remanding and vacating the D.C. Circuit decision would allow that court to consider the legal issue anew, taking into account the new administration’s “revised position.”

Intervenors—the State of New York plus three interested groups—urged the Court to deny the petition, defending DOE’s rule, explaining that the D.C. Circuit had appropriately interpreted the meaning of the statute, and arguing that the government’s change in position does not warrant a GVR. The intervenors pointed out that the administration had more than nine months to ask the D.C. Circuit to refrain from issuing a decision while the agency reconsidered its rule, but did not. The administration could have petitioned the D.C. Circuit for rehearing based on the errors it has now identified, but did not. And the agency could attempt a new rulemaking, but has not (DOE asserts that it could lawfully adopt a new, more lenient standard; intervenors argue that the statute’s anti-backsliding provision would prevent that). Intervenors also highlighted that Justices from across the Court’s ideological spectrum have recently interpreted GVRs as appropriate only in narrower circumstances than DOE has urged here.

DOE’s GVR request appears to be an effort to quickly clear the legal slate and adopt a contrary statutory interpretation that would allow a more lenient regulatory standard (and in the process avoid the effects of the statute’s anti-backsliding provision). Rather than opposing the industry petition, seeking abeyance, requesting rehearing, or initiating a new rulemaking, DOJ embraced the petition as an opportunity to urge the Court to vacate a recent decision upholding its own rule and reopen the statutory interpretation question.

That the government has taken a different position before the Supreme Court than before the D.C. Circuit is not on its own aberrant: confessions of error happen on average about twice per term, though they are more often based on mistakes than policy changes. And it is impossible from the filing alone to know whether the normal internal deliberative processes were followed. But this example—a policy-driven change in position, bypassing traditional tools, in service of a statutory interpretation that would curtail agency authority—fits with an identified pattern. The administration has repeatedly cited Loper Bright to argue that its preferred reading of a statute is the only permissible one, limiting agencies’ discretion and attempting to bind future administrations to that reading; DOE’s new interpretation of EPCA’s “performance characteristics” limitation appears to be consistent with that approach.

Separately, but importantly, the Court’s GVR order signals its openness to this kind of change in position. Historically, many members of the Court have expressed views that GVRs should be available only in limited circumstances, such as when there is intervening relevant precedent or confession of some material mistake. The leading treatise Supreme Court Practice (Shapiro et al.) characterizes GVRs based on the Solicitor General’s announced change in position as “controversial” and “the subject of substantial disagreement” on the Court.

Pre-Loper, many Justices saw GVR as potentially appropriate where the government announced a change in an agency’s statutory interpretation that was entitled to Chevron deference (since that change could influence the lower court’s resolution of the statutory interpretation question), though the Court was split on whether changed interpretations not entitled to deference might justify GVR. The 1996 majority and dissenting statements on GVR in Lawrence v. Chater illustrate the debate, with Justice Scalia arguing in dissent that the government “going back on what it argued to the court of appeals” was not enough to justify GVR where there was only a possibility that the government’s new statutory interpretation was entitled to deference. As the American Gas intervenors point out, many members of the Court have criticized the Lawrence majority’s approach. And after Loper, the considerations that may have persuaded the Court to GVR in Lawrence should not occur because courts would no longer defer to a changed agency interpretation as they might have under Chevron.

Moreover, the Court during the last administration appeared to be reluctant to grant GVRs based on changes in policy or interpretation, leading Justice Sotomayor to question whether the Court was “quietly constricting” the availability of GVRs based on intervening developments (at least in the criminal context). It remains to be seen whether the Court’s quick grant in American Gas is part of its own broader pattern of openness to using GVRs based on policy changes (and whether such a pattern holds across administrations and regardless of policy).

The case will now return to the D.C. Circuit, and it will be one to watch. Will DOE, through DOJ, ask the court to adopt its new legal view, to summarily vacate the DOE rule (which would run against a norm identified above), to hold the case in abeyance, or to do something else? And if DOE initiates a rulemaking to adopt a new legal interpretation, will it do so with a full participatory process or invoke some exception to attempt to speed its process?  

Erika B. Kranz is a staff attorney at the Environmental & Energy Law Program at Harvard Law School.