In re: Sealed Case

• Court: Court of Appeals for the D.C. Circuit
• Decided: July 18, 2025
• Docket: 24-5089
• Status: Published

Opinion

United States Court of Appeals FOR THE DISTRICT OF COLUMBIA CIRCUIT

Argued November 12, 2024 Decided July 18, 2025

No. 24-5089

IN RE: SEALED CASE

Appeal from the United States District Court for the District of Columbia (No. 1:24-mc-00007)

Ari Holtzblatt argued the cause for appellant. With him on the briefs was Jeremy W. Brinster.

Daniel J. Lenerz, Assistant U.S. Attorney, argued the cause for appellee. With him on the brief were Matthew M. Graves, U.S. Attorney, at the time the brief was filed, and Chrisellen R. Kolb, Assistant U.S. Attorney.

Before: MILLETT and GARCIA, Circuit Judges, and ROGERS, Senior Circuit Judge.

Opinion for the Court filed by Circuit Judge GARCIA.

GARCIA, Circuit Judge: The Stored Communications Act authorizes the government to subpoena electronic communication service providers for certain user records. The Act also permits the government to seek a court order prohibiting a service provider from disclosing that it received such a subpoena. A court may enter that order, however, only 2

“if it determines that there is reason to believe” disclosure “will” pose a safety risk or jeopardize the government’s investigation. 18 U.S.C. § 2705(b). This case involves a bold exercise of that statutory authority. The government asked for, and a court issued, an order allowing the government to prohibit the disclosure of any subpoena related to a particular investigation for one year, so long as the government decided that disclosure would risk one of the harms specified in the Act. The government then served a subpoena for user records on X Corp. and attached the nondisclosure order. X moved to vacate the nondisclosure order, arguing that it does not comply with the Act. The district court denied X’s motion. We reverse. The order does not conform to the Act because a court did not find “reason to believe” that disclosure “will” risk a statutory harm. In reaching that conclusion, however, we do not rule out the possibility that other prospective, multi- subpoena orders might satisfy the statute’s requirements. We do not reach X’s additional arguments that the order violates the First Amendment and that the district court improperly relied on ex parte evidence below. I A The Stored Communications Act protects the privacy of user data held by electronic communication service providers (like X), including by regulating government access to that data. See S. Rep. No. 99-541, at 3 (1986). To that end, the Act generally prohibits service providers from “divulg[ing]” the “contents” of any “communication” exchanged or stored on their platforms. 18 U.S.C. § 2702(a)(1)–(2); cf. id. § 2702(b) (describing “[e]xceptions”). It also generally prohibits providers from “divulg[ing]” user “record[s].” See id. § 2702(a)(3); cf. id. § 2702(c) (describing “[e]xceptions”). The Act does, however, permit release to the government of a 3

limited category of user records—such as biographical account information or associated IP addresses—pursuant to subpoenas and other legal process. See id. §§ 2702(c)(1), 2703(c)(2). Under the Act, government requests for user communications and user records are presumptively public. When the government seeks a user’s data, the service provider is typically allowed to notify that user (and the government is sometimes required to do so). See id. § 2703(b), (c). Section 2705(b) spells out an exception to that presumption. That provision authorizes the government to apply for a court order prohibiting providers from disclosing subpoenas, warrants, or court orders for customer records. Id. § 2705(b). A court shall issue such a nondisclosure order “if it determines that there is reason to believe” disclosure “will” prompt “(1) [danger to] physical safety . . . ; (2) flight from prosecution; (3) destruction of or tampering with evidence; (4) intimidation of potential witnesses; or (5) [other actions] seriously jeopardizing an investigation or unduly delaying a trial.” Id. B On December 11, 2023, the U.S. Attorney for the District of Columbia applied to a magistrate judge for a Section 2705(b) nondisclosure order. The government did not have any particular subpoena or subpoenas in hand when it submitted that application. Instead, it asked for authority to attach a nondisclosure order to any of the many subpoenas it might issue in one investigation over the next year. That same day, the magistrate judge granted the government’s request via two related orders. The first was a generic Section 2705(b) nondisclosure order commanding the unspecified “recipient of the attached subpoena” not to disclose the existence of that subpoena. J.A. 11. The second order— which we will call an “authorizing order”—outlined how the government was permitted to use that nondisclosure order. The 4

authorizing order permitted the government to attach the nondisclosure order to all subpoenas for user records issued in connection with a specified investigation over the next year. It also required the government to find, for “each and every subpoena to which the [nondisclosure o]rder is attached,” that “disclosure of the existence of th[at] subpoena would result in potential target(s) attempting to evade apprehension, or destroy or encrypt evidence, or otherwise seriously jeopardizing the investigation.” S.A. 2. The authorizing order thus effectively required the government to decide—before attaching the nondisclosure order to any given subpoena—that disclosure of the subpoena would risk one of the harms listed in Section 2705(b). On January 5, 2024, the government served X with a grand jury subpoena for records associated with two users’ profiles. The government attached the nondisclosure order to the subpoena, prohibiting X from disclosing the subpoena’s existence. X moved to vacate or modify the nondisclosure order, arguing that Section 2705(b) did not authorize the order and that the order violated the First Amendment. The district court found that the order was statutorily authorized and constitutional. In reaching its decision, the district court relied on evidence submitted by the government ex parte and largely declined X’s request that the government produce those ex parte submissions for X’s review. X appealed, challenging both the district court’s decision and its reliance on ex parte documents. II We have jurisdiction to review the district court’s denial of X’s motion to vacate or modify the nondisclosure order. To start, the district court’s denial was a final decision that we may review under 28 U.S.C. § 1291. A final decision “ends the litigation on the merits and leaves nothing for the court to 5

do but execute the judgment.” Quackenbush v. Allstate Ins. Co., 517 U.S. 706, 712 (1996) (citation modified). Here, the litigation centered on only one issue: the validity of the nondisclosure order. And the district court’s opinion conclusively resolved that issue, ruling that the order was permissible. No other action related to the court’s orders or the underlying subpoena is still pending before the district court. The district court’s denial thus ended the litigation on the merits and qualifies as a final decision.1 We also possess jurisdiction even though, as both parties agree, the case became moot while this appeal was pending. One month after X submitted its opening brief, a magistrate judge granted the government’s motion to vacate the nondisclosure order. X is thus free to disclose the grand jury subpoena, including to the users whose data the government had requested. It is therefore “impossible for [us] to grant any effectual relief” to X, and so the case is now moot. Decker v. Nw. Env’t Def. Ctr., 568 U.S. 597, 609 (2013) (citation omitted); see also In re Sealed Case, 77 F.4th 815, 826 (D.C. Cir. 2023). Both parties nonetheless contend that at least some aspects of this dispute fall within the “capable of repetition yet evading review” exception to mootness. We agree. Under that exception, a court may consider an otherwise- moot dispute where (1) “the challenged action” endured too briefly “to be fully litigated,” and (2) there is “a reasonable expectation that the same complaining party [will] be subjected to the same action again.” J. T. v. District of Columbia, 983

1 Our court relied on the collateral-order doctrine to justify review of a similar order in In re Sealed Case, 77 F.4th 815, 825–26 (D.C. Cir. 2023). The parties here agree that the collateral-order doctrine applies. As discussed above, however, this dispute is not collateral to any ongoing controversy between the parties, so it is reviewable without resort to the collateral-order doctrine. 6

F.3d 516, 523 (D.C. Cir. 2020) (citation omitted). This case meets both requirements. The court’s orders were in force for only six months and were set to expire after one year. Either period was too fleeting for litigation to run its course. See In re Sealed, 77 F.4th at 827 (noting that we have generally “considered an order of less than two years’ duration too short to be fully litigated” (citation modified)). X can also reasonably expect to face another nondisclosure order based on an authorizing order of this kind. X states that it “receives thousands of subpoenas, court orders, and warrants each year for [its] users’ information,” many of which are covered by nondisclosure orders. Appellant’s Brief 7–8; see also In re Sealed, 77 F.4th at 827 (deeming X’s claims regarding nondisclosure order capable of repetition, as X “previously has received, and challenged, nondisclosure orders attached to subpoenas”). The government acknowledges it is likely to again request broad authorization to serve nondisclosure orders like this one on service providers like X. The government’s brief, for example, agrees that much of this case is capable of repetition because “[i]t is sufficiently likely that the government will rely on multiple-subpoena, multiple-provider nondisclosure orders in the future, as it has in the past.” Appellee’s Brief xii. The government even calls this authorizing order “illustrative.” Id. at 39. We therefore have jurisdiction to decide whether a court may grant the government authority to use a nondisclosure order in the way it did here. Although we possess jurisdiction, the government contends that we cannot consider some of X’s claims for another reason: X needed (but lacked) a statutory cause of action to challenge the nondisclosure order on nonconstitutional grounds in district court. See id. at 31–32. The government did not raise this objection below and has offered no explanation for its failure to do so, so this argument is forfeited. See Keepseagle v. Perdue, 856 F.3d 1039, 1054 7

(D.C. Cir. 2017). Though we do not resolve the issue, we note our skepticism of the government’s position. On the government’s view, it seems that no one could ever ask a court to review a nondisclosure order for compliance with Section 2705(b). And the government identifies no analogous precedent suggesting that a cause of action is required under circumstances like these: where the government initiates a legal proceeding, the court issues an order in that proceeding, the government applies that court order against a party, and the party then asks the court to revisit its statutory authority to issue the order. We therefore proceed to the merits of X’s challenge. III The court did not comply with Section 2705(b) of the Stored Communications Act in this case. Section 2705(b) states that “[t]he court shall enter” a nondisclosure order only “if it determines that there is reason to believe that notification of the existence of the warrant, subpoena, or court order will result in” one of the statutorily specified harms. 18 U.S.C. § 2705(b). The statute’s text thus requires that a court make a “reason to believe” determination for any subpoena (or other legal document) covered by its order. See id. Neither party disputes this straightforward reading. Because the court must find “reason to believe” that disclosure of each covered subpoena “will result” in harm, the nature of the required analysis turns on the scope of the court’s order. If the order covers just one subpoena that is before the court, the inquiry is straightforward: The court must find reason to believe disclosing that subpoena will result in a statutory harm. If the order covers multiple subpoenas, though, the court must make a determination that addresses all those subpoenas. Similarly, if the order applies to future, hypothetical subpoenas, the court ordinarily must identify the types of subpoenas to which its order can apply so it can explain 8

why it has reason to believe disclosure of those potential subpoenas will cause harm. The authorizing order here had two unique features relevant to the required analysis: It permitted the government to apply the nondisclosure order prospectively to unidentified subpoenas, and to apply the nondisclosure order to subpoenas directed at a wide and unpredictable range of accounts. To comply with Section 2705(b), then, the court needed to make a “reason to believe” determination that accounted for a vast array of potentially covered subpoenas. Consider the authorizing order’s features, and their impact on the “reason to believe” analysis, in turn. First, the authorizing order applied prospectively, allowing the government to use the nondisclosure order for a full year. The magistrate judge thus needed to explain why she had “reason to believe” that disclosure of any covered subpoena potentially issued during the next year “will result” in harm. That determination differs from the analysis required to issue a nondisclosure order that can be attached only to an existing subpoena. For an existing subpoena, the magistrate judge can base her “reason to believe” determination on the facts as they were when the government submitted its application. But facts change. Over the next year, the public could learn of the government’s investigation to an extent that makes the nondisclosure order unnecessary. See, e.g., In re Sealed, 77 F.4th at 825. Or the target of the subpoena or investigation could be taken into custody or pass away, eliminating any risk that the target would destroy documents in their possession or flee upon learning of the investigation. As a result, to make the statutory finding, the court would need to give not only “reason to believe” that disclosure presently risks harm, but also “reason to believe” that risk of harm would still exist for a subpoena issued many months later. Second, the authorizing order provided no meaningful limit on the potential targets of the future subpoenas. It 9

permitted the government to attach the nondisclosure order to any subpoena for user records of any account with any service provider that it decided was relevant to its investigation. Thus, the government could attach the nondisclosure order not only to requests for the data of the primary targets of the investigation, but also (for example) to requests targeting those who played a merely peripheral role in the scheme. Moreover, because the authorizing order’s scope was dictated in part by the reach of the government’s investigation, it would be difficult for a court to predict whose accounts might be subpoenaed. After all, the government, not the court, directs the scope of a grand jury investigation. See United States v. Williams, 504 U.S. 36, 48 (1992). And the government sought this broad authority so that, as its investigation evolved, it could more quickly forbid disclosure of unforeseen subpoenas without coming back to the court. To satisfy the statute, then, the court needed to acknowledge the potential variety in the subpoenas to which the government could attach the nondisclosure order. Only then could it have given a “reason to believe” that disclosure of that full range of subpoenas “will result” in harm—including subpoenas that the government could not foresee when it requested this authority. Given these characteristics of the authorizing order, it would be exceedingly difficult for a court to reasonably predict what subpoenas might be covered, much less offer a “reason to believe” that disclosure of all those subpoenas “will result” in harm. In any event, the court certainly did not make the required determination here. The magistrate judge’s determination amounted to two lines: The judge found “reasonable grounds to believe that disclosure of such subpoenas will result in flight from prosecution, destruction of or tampering with evidence, intimidation of potential witnesses, and serious jeopardy to the investigation.” S.A. 1. And issuance of an “omnibus Order” was “warranted,” she said, “based on the nature of the criminal 10

activity, the volume of expected subpoenas, and the burden that separate applications would pose.” Id. That analysis does not suffice. The magistrate judge did not address how the court’s “reason to believe” applied to subpoenas that might be issued months in the future. Nothing suggests that the judge grappled with the potential variety in the subpoenas, including the range of targets, that her orders might cover. And the judge never explained how her “reason to believe” applied to each of those subpoenas. The government has little to say in response. It does not argue that the magistrate judge made the findings the statute requires. The government instead says that the orders were validly issued because the court required the government to make those findings. The government emphasizes the authorization order’s requirement that, before the nondisclosure order could be attached to a future-issued subpoena, the government needed to evaluate whether “disclosure of the existence of the subpoena would result in potential target(s) attempting to evade apprehension, or destroy or encrypt evidence, or otherwise seriously jeopardizing the investigation.” S.A. 2. But that provision only confirms the statutory violation. The statute requires “[t]he court” to “determine[]” that disclosure of the subpoena’s existence “will result in” harm. 18 U.S.C. § 2705(b) (emphasis added). The authorizing order, though, seems to have outsourced to the government the very evaluation that Congress assigned to the court. In sum, the court was required to find reason to believe that a harm enumerated in Section 2705(b) will arise from disclosure of any subpoena potentially covered by the nondisclosure order. We have no occasion to provide a precise blueprint for what form those findings must take when a court prospectively authorizes use of a nondisclosure order with future subpoenas, as it did here. But to find that disclosure of 11

each subpoena (or perhaps appropriate categories thereof) will result in a Section 2705(b) harm, a court must at least match its required statutory findings to the breadth and variety of potential subpoenas to which its nondisclosure order could apply. Here, we hold that the court did not make the required finding before issuing these orders and thus did not conform to Section 2705(b). Our holding, however, is narrow. Although X has raised substantial arguments that the statute bars all orders purporting to prospectively authorize the government to attach a nondisclosure order to unissued subpoenas, we need not resolve those categorical arguments here. To be sure, we are skeptical that a court could intelligibly make the determination Section 2705(b) requires for any prospective order as broad as this one. But we cannot rule out the possibility that, in another case, the government might seek a prospective order with clearer limitations or other features that could allow a court to make the requisite findings. We also clarify one additional point. Much of the parties’ briefing focused on whether Section 2705(b) permits the government to seek a single nondisclosure order that could be attached to multiple subpoenas. As a statutory matter, it is not inherently problematic for a court to use a single order to mandate nondisclosure of multiple subpoenas. As X concedes in its reply brief, an “order that analyzed and applied to multiple, existing subpoenas presented to the magistrate could be lawful.” Reply Brief 5. This nondisclosure order violates the statute not because it applies to multiple subpoenas, but because the court failed to make the required “reason to believe” determination for those subpoenas. IV In light of our statutory ruling, we decline to reach X’s alternative and independent argument challenging the nondisclosure order on First Amendment grounds. See In re 12

Leopold, 964 F.3d 1121, 1126–27 (D.C. Cir. 2020) (noting that, per “longstanding principle[s] of judicial restraint,” we “avoid unnecessarily passing on a constitutional question of first impression” where appellants “can receive all of the relief they request” on alternative grounds (citation modified)). We also do not reach X’s claim that the district court improperly relied on ex parte materials in upholding the order. Our statutory ruling renders that dispute irrelevant. The order was invalid regardless of whether it was lawful for the district court to rely on ex parte evidence. V The district court’s judgment is reversed. So ordered.