The Supreme Court’s recent decision in FCC v. AT&T might restore agency control over initial adjudications that many thought Jarkesy to forbid. As Jeff Wall (counsel for AT&T) argued, this case lets the government “drag in through the back door what in Jarkesy said you couldn’t bring through the front.” Transcript of Oral Argument at 7-8, FCC v. AT&T, No. 25–406, slip op. (2026). 

The case presented a straightforward question: does the Seventh Amendment permit the Federal Communications Commission to adjudicate a $57 million forfeiture penalty against AT&T without a jury? Two years after SEC v. Jarkesy declared that agencies cannot impose civil penalties in-house without a jury trial, the Court answered yes, 8-1. The opinion is written as a narrow distinction on the grounds that the FCC’s order has no technical legal effect on questions of law or fact. Read carefully, however, FCC v. AT&T supplies the template agencies need to replicate the very adjudications Jarkesy was supposed to prohibit.

To understand what AT&T gives away, start with what Jarkesy won. Before Jarkesy, the Court’s public rights case law evolved to permit agencies to adjudicate more and more types of claims. See, e.g., Crowell v. Benson, 285 U.S. 22 (1932) (permitting agency adjudication of private rights); FTC v. Schor, 478 U.S. 833(1986) (permitting agency adjudication of state law claims); Wellness Int’l Network, Ltd. v. Sharif, 575 U.S. 665 (2015) (permitting bankruptcy court adjudications over non-bankruptcy claims with consent of parties). But more recently, the Court’s Seventh Amendment case law began restricting the public rights exception wherever parties were entitled to a jury trial. See Granfinanciera, S.A. v. Nordberg, 492 U.S. 33 (1989). And in Jarkesy, the Court introduced a new framework for evaluating whether agencies can initially adjudicate claims that appeared to seriously rollback agency adjudication over a whole swath of claims.

Jarkesy laid out a two-part framework for assessing whether a claim can be initially adjudicated in an agency. SEC v. Jarkesy, 603 U.S. 109 (2024). First, the Court asks whether the Seventh Amendment is implicated, which turns on “the nature of the claim” and whether the “remedy is . . . designed to punish or deter the wrongdoer.” Id. at 123. Second, if the Seventh Amendment is implicated, courts ask whether the public rights exception applies. Id. at 127. That question turns on whether the claim is “made of the stuff of the traditional actions at common law” or belongs to “a class of cases” that “historically could have been determined exclusively by the executive and legislative branches.” Id. at 128 (cleaned up). 

AT&T focuses on Jarkesy’s first prong. Specifically, it addresses when agency action–regardless of the type of claim–counts as a “suit at common law” that triggers the Seventh Amendment to begin with. U.S. CONST. amend. VII. In Jarkesy, the penalties at issue were immediately enforceable by the SEC. Id. at 118. The moment the SEC issued a final order, it could garnish the recipient’s wages and deduct penalties. If the SEC chose to go to court, no jury was available on the underlying merits—courts at that stage were merely enforcing an already-final agency determination. The agency both adjudicated and collected, without an Article III court ever resolving the merits initially. Thus, the agency’s action constituted a suit at common law. Not so in AT&T.  

The Distinguishing Move in AT&T

In AT&T, the FCC’s enforcement scheme, the Court held, was meaningfully different than in Jarkesy. Under the Communications Act, the FCC may adjudicate forfeiture penalties through an informal proceeding: it issues a notice of apparent liability, reviews the carrier’s response, and then “determines,” “assesses,” and “imposes” a penalty. FCC v. AT&T, at 2. But the Commission cannot execute on its own order. It cannot seize assets, obtain liens, or garnish wages. Interest does not accrue on unpaid penalties. There are no sanctions for nonpayment. And under 47 U.S.C. § 504(c), the Commission cannot even cite an unresolved forfeiture order against the carrier in subsequent proceedings. Id. at 7-8. If the government wants to collect, the Department of Justice must bring a civil suit, and that suit “shall be a trial de novo.” Id. at 3 (quoting 47 U.S.C. § 504(a)). For purposes of that trial, the Court explained, “it is as if the Commission never found any facts at all.” Id. at 8.

The Jarkesy distinction, so framed, is coherent. Where the SEC both adjudicated and enforced its penalty without any jury ever resolving the merits, the FCC’s findings carry no legal weight until a jury independently confirms them in a subsequent enforcement action in federal court. The forfeiture order, on this view, is not a determination of rights; it is a “prerequisite to suit,” analogous to a right-to-sue letter, that enables the DOJ to bring a collection action. Id. at 11. The Seventh Amendment does not attach to preliminary procedures, so the Court reasons.

The statute does give carriers one alternative to waiting for a DOJ enforcement suit: pay the penalty and seek review directly in the court of appeals under the Hobbs Act. Id. at 3. The Court itself acknowledges the problem with that path, albeit in a footnote. See id. at 12, n. 4. Hobbs Act review in the court of appeals applies APA standards—deferential review on the administrative record, not a de novo trial. Id. And the Court expressly declines to decide whether payment is even a prerequisite to accessing that review at all. Id. So the carrier’s practical choice reduces to this: pay over $50 million for the privilege of deferential appellate review, or decline to pay and wait indefinitely for the DOJ to bring a de novo suit it has no obligation to file. As explained below, it is not clear that a jury trial under these circumstances would fairly qualify as a de novo trial.

The De Novo Mirage

The Court’s entire constitutional architecture (that there is no Seventh Amendment jury trial deprivation) rests on one load-bearing assumption: that AT&T would get a genuine de novo trial even after the agency issues its order. That assumption drives a hole through Jarkesy for two reasons. First, when courts do de novo review of legal questions in practice, their analysis often reduces to deference towards the agency’s legal conclusions. Second, a robust pre-trial agency record will most likely influence the court if not control its factual conclusions. 

Begin with the likelihood of deference even under a de novo standard. The Court states that the “trial de novo includes de novo review of the Commission’s legal conclusions.” AT&T, at 8, n.1 (citing McLaughlin Chiropractic Associates, Inc. v. McKesson Corp., 606 U.S. 146, 152 (2025)). But in the next sentence, the Court acknowledges that multiple circuits have held, in § 504 enforcement actions, that district courts are precluded from reviewing the Commission’s legal conclusions in § 504 suits. Id. at 8, n.2 (citing United States v. Stevens, 691 F.3d 620, 622 (5th Cir. 2012). In other words, the standard of review is irrelevant for AT&T if circuit precedent precludes review of law in the first place under any standard. In those circuits, courts would be compelled to defer to the Commission’s legal conclusions. But even in the other federal circuits, courts can still say that the “best meaning” of the Communications Act is that courts should almost always defer to the learned conclusions of expert bodies like the FCC. Indeed, the Court has expressly affirmed this type of reasoning in recent cases like Seven County Infrastructure Coalition v. Eagle County, where the Court’s “de novo” review of NEPA standards yielded deference to agency judgments. 605 U.S. 168 (2025).

Turn next to the influence that robust pre-trial records will have on the actual trial. Chief Justice Marshall identified this problem two centuries ago in Osborn v. Bank of the United States, 22 U.S. 738 (1824). A litigant’s constitutional right to a federal tribunal, Marshall wrote, is made hollow when he is “restricted to the insecure remedy of an appeal upon an insulated point, after it has received that shape which may be given to it by another tribunal, into which he is forced against his will.” Id. at 822. A court confronting the FCC’s prior determination—detailed factual findings, legal analysis, a specific dollar figure—does not and cannot begin from a blank slate. The agency has already “given shape” to the case. Justice Frankfurter recognized the same dynamic from the opposite direction in Universal Camera Corp. v. NLRB, insisting that courts reviewing agency factual findings examine “the record considered as a whole” precisely because the tendency to treat the agency’s account as the baseline—and look only for evidence supporting it—was already well-established even under formal substantial evidence review. 340 U.S. 474 (1951). If that tendency appears under a standard of review that is already deferential, there is every reason to expect it to operate at least as powerfully in a de novo proceeding where the agency’s prior work is the starting point.

Thus, even if AT&T availed itself of a jury trial (assuming the DOJ brought suit), the Court’s promise of de novo review of both law and facts is an illusory one.

The Broader AdLaw Picture

The scope of the Court’s holding in AT&T could be sweeping. At oral argument, AT&T’s counsel pointed out that Congress deployed the same mandatory language—“assesses,” “imposes,” “shall be liable”—in the FCC statute as in the SEC statute the Court struck down in Jarkesy. See Transcript of Oral Argument at 5, FCC v. AT&T, No. 25–406, slip op. (2026). The majority’s response was that this language, read “alone and in isolation,” cannot establish whether an order truly settles legal rights. AT&T, at 9 (cleaned up). Looking to “the overall statutory scheme,” the Court held that this statutory language vested no power in the FCC to “visit any adverse consequences on a regulated party who receives a forfeiture order.” Id. at 9-10. The implication is that all other statutes that similarly appeared to confer immediate enforcement power on agencies also only create the power to issue non-binding agency opinion documents with no legal effect. All this despite that the government has “for 50 years . . . understood these orders to impose a present and binding legal obligation to pay.” Transcript of Oral Argument at 94-95, FCC v. AT&T, No. 25–406, slip op. (2026).

But perhaps this case reflects a broader trend in the administrative law arc of the Roberts Court. The conventional wisdom holds that this Court has been aggressively restricting the administrative state: Chevron overruled (Loper Bright), the nondelegation doctrine stirring (at least before Consumers Research), agency adjudications constitutionally suspect (Jarkesy). But in landmark case after landmark case, the Court leaves an opening just wide enough for the next case to drive through. Loper Bright commanded courts to exercise independent judgment on questions of statutory interpretation, and then noted that the “best” reading of ambiguous text can itself be that Congress delegated discretion to the agency. Seven County arrived the following term and held that agency determinations about the adequacy of environmental review are subject only to arbitrariness review, not independent judgment. Bondi v. VanDerStok suggested that even where a party brings a pre-enforcement facial challenge to an agency rule subject to de novo review, only as-applied challenges are likely viable. The pattern is consistent: the Court announces a structural limit on agency power, then supplies, in the same breath or the next case, the doctrinal escape hatch.

Jarkesy proves no different. Jarkesy created a clean, structural rule that initial adjudications of common-law-style claims belong in Article III courts, not agency tribunals. After FCC v. AT&T, an agency can open a formal proceeding, receive written submissions, weigh the evidence, interpret the governing statute, and declare in capital letters that a party “IS LIABLE TO THE UNITED STATES FOR A MONETARY FORFEITURE” of $57 million—and do all of it without a jury, as long as the order is technically unenforceable without a follow-on DOJ lawsuit that may never come. See AT&T, at 3 (Thomas, J., dissenting). If it walks like a duck and quacks like a duck, the Roberts Court instructs that it is not a duck, provided the duck lacks the power to collect its own eggs. FCC v. AT&T should therefore surprise no one. It reflects a Court that strikes fast and decisively in the landmark and then, case by case, sands down the edges until what remains is a principle that announces much and demands little.

Ryan Keane is a 2026 graduate of Harvard Law School and an incoming law clerk on the Western District of Texas and later on the Seventh Circuit.