Coalition for Independent Technology Research v. Rubio

Opinion

UNITED STATES DISTRICT COURT FOR THE DISTRICT OF COLUMBIA

COALITION FOR INDEPENDENT TECHNOLOGY RESEARCH,

Plaintiff,

v. Civil Action No. 26-815 (JEB)

MARCO RUBIO, et al., Defendants.

MEMORANDUM OPINION

The town square of yesteryear has moved online. Where public debate once played out on street corners and in the daily papers, on soapboxes and the evening news, it now unfolds in large part on a handful of internet platforms owned and operated by private entities. Behind the posts, feeds, labels, and takedowns that shape what users see is a sprawling ecosystem of platforms, researchers, fact checkers, advocates, and trust-and-safety professionals. Some study how false or harmful content spreads, others press platforms to change their rules, and still others help users respond to online abuse. To one side of a heated public debate, that work makes digital discourse safer and more accountable. To the other, it is censorship by another name.

The State Department has now placed immigration consequences behind that latter view.

In 2025, the Secretary announced a policy targeting foreign nationals said to be complicit in censoring Americans. What began as a visa-restriction policy later expanded, according to Plaintiff Coalition for Independent Technology Research, into a broader campaign against noncitizens who work on misinformation, disinformation, fact checking, content moderation, compliance, and trust and safety. The Department has since invoked that policy to bar

2

individuals from the country or seek their removal, including leaders of CITR member organizations.

In response, CITR has brought this action, and it now seeks both a preliminary injunction and a stay of the policy. It contends that the policy chills its members’ research, advocacy, travel, and association and, in turn, impairs CITR’s own reporting, convening, and public-facing work. The Court concludes that Plaintiff has shown a likelihood that the policy is reviewable and that it burdens protected speech and association on the basis of viewpoint, in violation of the First Amendment and the Administrative Procedure Act. As the remaining preliminary-relief factors also favor Plaintiff, the Court will grant a stay of the policy under 5 U.S.C. § 705.

I. Background A. Statutory Framework Two provisions of the Immigration and Nationality Act set the table. The first governs entry. A noncitizen is inadmissible if the Secretary of State has “reasonable ground to believe” that the individual’s “entry or proposed activities in the United States . . . would have potentially serious adverse foreign policy consequences for the United States.” 8 U.S.C. § 1182(a)(3)(C)(i). That broad authority is not without limitations. A noncitizen “shall not be excludable or subject to restrictions or conditions on entry into the United States” based on that individual’s “past, current, or expected beliefs, statements, or associations, if such beliefs, statements, or associations would be lawful within the United States.” Id., § 1182(a)(3)(C)(iii). That hedge, however, has limitations of its own. The Secretary may exclude noncitizens based on protected “beliefs, statements, or associations” if he “personally determines that the alien’s admission would compromise a compelling United States foreign policy interest.” Id. Should he so determine, the Secretary “must notify on a timely basis” the Judiciary and Foreign Affairs

3

Committees of the House and the Judiciary and Foreign Relations Committees of the Senate, identifying the noncitizen and the justification for his exclusion. Id., § 1182(a)(3)(C)(iv).

The second provision concerns removal. A noncitizen already in the country “is deportable” on much the same showing as what governs admission: the Secretary of State determines that the noncitizen’s “presence or activities . . . would have potentially serious adverse foreign policy consequences for the United States.” Id., § 1227(a)(4)(C)(i). As with inadmissibility, the deportation provision shields noncitizens from removal based on lawful beliefs, statements, or associations. Id., § 1227(a)(4)(C)(ii) (incorporating id., § 1182(a)(3)(C)(iii)). The Secretary, once more, may override that protection by making an individualized determination that the noncitizen’s presence compromises a compelling foreign- policy interest. Id.

B. Factual Background 1. Content Moderation Online platforms now carry an outsized share of our political arguments, exchange of critical news, and commerce, as well as much of our private conversations. This migration away from traditional forms of media has recast a timeless question: how, and by whom, should the bounds of public debate be drawn? No online platform desires to host all that its users might post, and so each must decide what to carry and what to refuse. Those decisions are not made once but continuously, across millions of messages, images, and videos that are uploaded each day. A platform may let a post stand, it may bury it far down a feed, it may attach a label cautioning that claims are unverified or conjured by artificial intelligence, or it may take down a post altogether.

4

Meta, for example, stepped back from centralized enforcement by swapping “third-party fact checking” in the United States for reader-written “Community Notes” and trimming speech- restriction rules. See More Speech and Fewer Mistakes, Meta (Jan. 7, 2025), https://perma.cc/U5ZZ-U8JQ. X takes a more permissive approach, letting even graphic media stand so long as it is labeled and not prominently displayed. See The X Rules, X, https://perma.cc/LL2U-LY6D. Bluesky declines the role of sole arbiter altogether, setting a baseline and then letting users subscribe to independent, stackable “labeling” services that decide much of what each person sees hidden, blurred, or flagged. See Bluesky’s Stackable Approach to Moderation, Bluesky (Mar. 12, 2024), https://perma.cc/PVY5-PWWM. This never-ending determination about what information a service will carry, and on what terms, is the essence of content moderation.

While such practice is easy to describe, it is anything but settled. Public debate over the subject is active and divided. See, e.g., Christopher St. Aubin & Jacob Liedke, Most Americans Favor Restrictions on False Information, Violent Content Online, Pew Rsch. Ctr. (July 20, 2023), https://perma.cc/VSQ3-RHFA (reporting growing “partisan gap in support for restricting false information”). To one camp, the platforms moderate far too much, suppressing lawful speech or disfavored political speech under the banner of combating falsehood and hate. To another, they moderate too little, letting lies, harassment, and incitement spread and harden into real-world harm. The core of the dispute is whether a platform’s decision to bury or delete a post is the responsible tending of the town square or the heavy-handed silencing of a view. One person’s content moderation, in the end, is another’s censorship.

The debate is not merely philosophical. An early marker came in 2020. After a platform appended a fact check to a post by the President addressing mail-in voting, he issued an

5

executive order condemning platforms’ “selective censorship” and directing agencies such as the Department of Justice to “review the viewpoint-based speech restrictions imposed by” those platforms. See Exec. Order No. 13925, 85 Fed. Reg. 34079, 34079, 34081 (June 2, 2020).

With a change in administrations in 2021, the effort passed largely to Congress, whose sights moved from the platforms to the researchers who studied them. A House subcommittee, for instance, subpoenaed the Stanford Internet Observatory — a university program devoted to the study of disinformation — on the theory that its research was itself “the censorship of disfavored speech.” Press Release, H. Comm. on the Judiciary, Chairman Jordan Presses Stanford on Subpoena Compliance for Censorship Investigation (June 1, 2023), https://perma.cc/99VS-KSYN. Under the weight of that inquiry, broad document demands, and private litigation, the Observatory lost its funding and leadership and ultimately wound down. See Joseph Menn, Stanford’s Top Disinformation Research Group Collapses Under Pressure, Wash. Post (June 14, 2024), https://perma.cc/G3RE-B7Y4. The subcommittee did not stop there, pressing similar demands on other researchers and advocacy organizations, among them members of CITR, and reaching even to regulators abroad. See ECF No. 1 (Compl.), ¶ 33.

Returning to office in 2025, President Trump once again turned his attention to the matter. An executive order issued in January of that year declared it the policy of the United States to end “censorship” and recast the prior administration’s efforts to “combat[] ‘misinformation,’ ‘disinformation,’ and ‘malinformation’” as attempts to pressure platforms “to moderate, deplatform, or otherwise suppress speech.” Exec. Order No. 14149, 90 Fed. Reg. 8243, 8243 (Jan. 28, 2025). A related order issued the same month directed the Secretary of State, Attorney General, Secretary of Homeland Security, and Director of National Intelligence to “[r]ecommend any actions necessary to protect the American people from the actions of

6

foreign nationals who have undermined or seek to undermine the fundamental constitutional rights of the American people, including, but not limited to, our Citizens’ rights to freedom of speech.” Exec. Order No. 14161, 90 Fed. Reg. 8451, 8452 (Jan. 30, 2025).

2. The Policy

Secretary of State Marco Rubio took up the invitation. On May 28, 2025, he announced “a new visa restriction policy that will apply to foreign officials who are responsible for censorship of protected expression in the United States.” Press Release, U.S. Dep’t of State, Announcement of a Visa Restriction Policy Targeting Foreign Nationals Who Censor Americans (May 28, 2025), https://perma.cc/3PBU-J257 (May 28 Press Release). An internal memorandum approved five days earlier set the groundwork for the policy, invoking the inadmissibility grounds of 8 U.S.C. § 1182(a)(3)(C) and citing the need “to secure the right of the American people to engage in constitutionally protected speech.” ECF No. 45-2 (May Mem.) at ECF pp. 1–2. The policy authorized the State Department to “restrict visa issuance” not only for “foreign nationals who are responsible for, or complicit in, censorship or attempted censorship of protected expression” but also for “the immediate family members of such persons.” Id. at ECF p. 1. Examples offered for being “responsible for, or complicit in, censorship” included the following: “threatening arrest for activity on American tech platforms,” fining platforms, detaining platforms’ employees “for not complying with . . . content moderation or censorship demands,” and demanding access to private user data to ensure compliance with content moderation policies. Id. at ECF p. 2.

The Department first put the policy to use abroad in July 2025. Casting Brazilian proceedings against former President Jair Bolsonaro as a “political witch hunt” that “created a persecution and censorship complex,” it revoked the visas of a justice of Brazil’s Supreme

7

Federal Court along with those of several of his colleagues and family members. See Press Release, U.S. Dep’t of State, Announcement of Visa Restrictions on Brazilian Judicial Officials and Their Immediate Family Members (July 18, 2025), https://perma.cc/3WWN-LCSB.

The policy’s reach widened that December. An internal cable — reported in the press and acknowledged by the Government, though never released — invoked the May policy to establish new criteria for H-1B visas typically used by tech companies. See ECF Nos. 11-15 (NPR Article) at ECF pp. 2–4; 50 (Hr’g Tr.) at 22:21–24. As part of those criteria, consular officers were instructed to “thoroughly explore” applicants’ “work histories,” “resumes, LinkedIn profiles,” and media appearances for involvement in “combatting misinformation, disinformation or false narratives, fact-checking, content moderation, compliance, and trust and safety,” and, upon discovering such involvement, to “pursue a finding that the applicant is ineligible for a visa.” NPR Article at ECF pp. 3–4 (quotation marks omitted).

Enforcement actions that followed the May policy went a step beyond holding the line at the border. The catalyst came from abroad: on December 5, the European Commission fined the platform X roughly $140 million under the Digital Services Act for breaching platform- transparency rules, see ECF No. 11-21 (EU Fine Announcement) at ECF pp. 2–3, a penalty the Secretary deemed “an attack on all American tech platforms and the American people.” ECF No. 11-22 (Rubio X Post of Dec. 5, 2025). As a response, Under Secretary Sarah B. Rogers sent the Secretary an action memorandum recommending that, “[c]onsistent with those foreign policy objectives articulated in the May 2025 3C policy,” the Secretary find two lawful permanent residents deportable. See ECF No. 11-23 (Dec. Action Mem.) at 2, 5–6. On December 19, the Secretary did so as to Imran Ahmed, the CEO of the Center for Countering Digital Hate

8

(CCDH), determining “based on the same foreign policy objectives articulated in [the] May announcement” that Ahmed was deportable under 8 U.S.C. § 1227(a)(4)(C). Id. at 7.

Four days later, the Secretary announced action against five individuals he described as having “led organized efforts to coerce American platforms to censor, demonetize, and suppress American viewpoints they oppose.” ECF No. 11-24 (Enforcement Action Press Release) at ECF p. 2. He barred all five from the country, noted that the Department of Homeland Security could begin removal proceedings as needed on the strength of his determinations, and warned that the Department “stands ready and willing to expand” the list. Id.

What the five had in common was not office or authority but work in the space of content moderation. Thierry Breton, a former European Commissioner, was named for a letter he had sent X owner Elon Musk in 2024, warning that the platform would face penalties under European law if it did not curb illegal content. See Compl., ¶ 56. Ahmed was named for CCDH’s research documenting hate and disinformation on the platforms and its campaigns pressing advertisers and the platforms to act on what it found. Id., ¶ 57. Clare Melford was named for the work of the Global Disinformation Index (GDI), which at the time published “disinformation risk ratings” for news and information sites. Id., ¶ 58. And Josephine Ballon and Anna-Lena von Hodenberg were named for their work at HateAid, a German nonprofit that helps targets of online abuse seek removal of, and legal redress for, the content aimed at them. Id., ¶¶ 59–60. Ahmed — who was living in the United States as a lawful permanent resident — received no advance notice of his deportability determination but secured a temporary restraining order against his detention in a separate suit. See Ahmed v. Rubio, No. 25-10705, ECF No. 14 (Order) (S.D.N.Y. Dec. 25, 2025).

9

The Department was not finished. Weeks later, in its 2026–2030 strategic plan, it pledged to counter such “censorship” of Americans “through all appropriate means including visa and financial sanctions.” Compl., ¶ 63.

3. The Coalition for Independent Technology Research CITR is a nonprofit membership organization formed in 2022 after Meta demanded that two New York University researchers abandon their study of how disinformation spreads on its service. Id., ¶ 122. Its mission is to “advance, sustain, and defend the right to ethically study the impact of technology on society,” free of the “obstruction, interference, and cooption” of governments and platforms alike. Id.; ECF No. 11-2 (Declaration of Brandi Geurkink), ¶ 4. Two programs carry out that mission. Through “Community,” CITR builds and sustains a network of support for researchers, fact checkers, and others working in the field, connecting them across institutions and borders, convening them at workshops and a yearly summit, and giving them the peer relationships and shared resources the work depends on. See Compl., ¶ 124. Through “Advocacy,” it conducts public-education campaigns and publishes its own research, including reports on the state of independent technology research. Id., ¶¶ 125, 135. To date, CITR counts roughly 500 individual and organizational members across 47 countries. Id., ¶ 123.

For CITR and its members, the policy looms large. The enforcement actions against Ahmed and Melford are the first illustration: each leads an organization — CCDH and GDI respectively — that is a member of CITR. Id., ¶¶ 57–58. But the two are not alone in their exposure. CITR’s members are researchers, fact checkers, and advocates whose work focuses on “misinformation, disinformation, content moderation, and trust and safety issues” — in other words, the content moderation that others see as censorship. Id., ¶ 66.

10

That reach has reshaped the Coalition and the work of its members alike. Fearing that travel to or from the United States will draw scrutiny, some of CITR’s noncitizen members have narrowed their research, declined to speak publicly or attend conferences, and pulled back from the Coalition’s work; a few have made plans to leave the country altogether. Id., ¶¶ 9, 65. CITR itself has suspended its in-person events in the United States, planned an additional 2026 summit at considerable cost to protect international members, watched members withhold their contributions from its own reporting, and diverted staff from its initiatives to answer members’ fears about the policy. See Geurkink Decl., ¶¶ 15–33; Compl., ¶¶ 128, 130, 135–136, 138. Nor have the effects stopped at the noncitizens the policy targets; CITR’s U.S.-citizen members have felt it, too, as the colleagues and collaborators they depend on retreat from shared work. See Compl., ¶¶ 108–20.

C. Procedural Background Plaintiff filed this action on March 9, 2026, against Rubio, then-Secretary of Homeland Security Kristi Noem, and then-Attorney General Pamela Bondi in their official capacities. Id., ¶¶ 16–18. The Complaint pleads three counts. First, CITR alleges that State’s policy contravenes the First Amendment by discriminating against viewpoints the Government has come to disfavor and that Defendants’ threats to strip visas and pursue removal for protected work amount to an impermissible campaign of coercion. Id., ¶¶ 139–41. Second, the organization contends that the policy is void for vagueness under the Fifth Amendment. Id., ¶¶ 142–45. Finally, CITR asserts that the policy violates the Administrative Procedure Act as agency action that is contrary to constitutional right, arbitrary and capricious, and in excess of the Secretary’s authority under the INA. Id., ¶¶ 146–48.

11

Later that month, CITR moved to stay the policy under 5 U.S.C. § 705 and to preliminarily enjoin Defendants from enforcing it during the pendency of this litigation. See ECF No. 11-1 (Pl. Mot.). The Court heard oral argument on the opposed Motion, see ECF No. 45 (Opp.), on May 13. See Minute Entry of May 13, 2026. II. Legal Standard Requests for relief under 5 U.S.C. § 705 are evaluated under the same standards as requests for preliminary injunctions. Cuomo v. NRC, 772 F.2d 972, 974 (D.C. Cir. 1985). Although the two remedies differ in operation — with an injunction commanding a party and a stay suspending the agency action itself — the inquiry that precedes them does not. See Nken v. Holder, 556 U.S. 418, 428–29 (2009).

“A preliminary injunction is an extraordinary remedy never awarded as of right.” Winter v. NRDC, 555 U.S. 7, 24 (2008). “A plaintiff seeking a preliminary injunction must establish [1] that he is likely to succeed on the merits, [2] that he is likely to suffer irreparable harm in the absence of preliminary relief, [3] that the balance of equities tips in his favor, and [4] that an injunction is in the public interest.” Sherley v. Sebelius, 644 F.3d 388, 392 (D.C. Cir. 2011) (alterations in original) (quoting Winter, 555 U.S. at 20). “[I]f the government is the opposing party,” the latter two factors “merge.” Glob. Health Council v. Trump, 153 F.4th 1, 12 (D.C. Cir. 2025).

“The moving party bears the burden of persuasion and must demonstrate, ‘by a clear showing,’ that the requested relief is warranted.” Hosp. Staffing Sols., LLC v. Reyes, 736 F. Supp. 2d 192, 197 (D.D.C. 2010) (quoting Chaplaincy of Full Gospel Churches v. England, 454 F.3d 290, 297 (D.C. Cir. 2006)). A court “may deny a motion for preliminary injunction, without further inquiry, upon finding that a plaintiff is unable to show either irreparable injury or

12

a likelihood of success on the merits,” making those two factors “particularly crucial.” Luokung Tech. Corp. v. Dep’t of Def., 538 F. Supp. 3d 174, 182 (D.D.C. 2021) (emphasis in original) (quotation marks omitted). III. Analysis The Court addresses each of the four preliminary-injunction factors in turn, devoting the bulk of its attention to CITR’s likelihood of success on the merits.

A. Likelihood of Success on Merits The likelihood-of-success inquiry here presents not one question but several, and their order matters, for each turns in part on the answers that precede it. The Court begins with the nature of the policy CITR challenges — whether it is the unified course of agency conduct CITR describes or the visa-only measure severed from any determination of removal. It then takes up the two threshold barriers the Government raises before the merits: CITR’s standing to sue and a set of jurisdictional bars in the INA that Defendants believe forecloses review of the policy and the relief CITR seeks. After clearing all of this substantial brush, the Court reaches the merits of CITR’s claims under the First Amendment and the Administrative Procedure Act. As those suffice to establish a likelihood of success, it need not address CITR’s Fifth Amendment vagueness claim.

1. Nature of Policy

The parties’ first disagreement is not about standing, jurisdiction, or the merits, but rather the object to which all three attach: the policy itself. CITR contends that it has challenged a single, unified course of agency conduct that both threatens noncitizens’ visas and renders them deportable for research and advocacy efforts that are branded as censorship. See Pl. Mot. at 4–9. Defendants rejoin that there is no such unified policy, but only a narrow visa-restriction measure

13

set out in May 2025, accompanied by separate and individualized removal determinations. See Opp. at 3–6. Severed that way, the Government’s conduct would present not a policy but a set of individualized decisions, the kind channeled elsewhere and heavily deferred to. As the resolution of this debate shapes much of what follows, the Court takes it up first.

The State Department indisputably announced a visa-restriction policy under 8 U.S.C.

§ 1182(a)(3)(C) in May 2025. See May 28 Press Release. The question, then, is whether it later applied that announced policy to a wider set of individuals engaged in content moderation, or whether those December actions were separate and unconnected measures. The answer lies principally in how the Department itself described the December actions. Those contemporaneous descriptions are not conclusive merely because the Department offered them, see Gen. Motors Corp. v. Ruckelshaus, 742 F.2d 1561, 1564–65 (D.C. Cir. 1984) (en banc), but they are probative of whether officials responsible for the December actions understood themselves at the time to be implementing the May policy. Cf. Williams Gas Processing — Gulf Coast Co. v. FERC, 373 F.3d 1335, 1345 (D.C. Cir. 2004) (agency action is measured by “reasoning that is fairly stated by the agency” at the time, not “post hoc rationalizations by agency counsel”) (citation omitted).

The May Memo, approved on May 23, 2025, and announced five days later, invokes the inadmissibility ground of 8 U.S.C. § 1182(a)(3)(C) and authorizes the Department to “restrict visa issuance for foreign nationals who are responsible for, or complicit in,” censorship of protected expression, together with their immediate family members. See May Mem. at ECF p. 1. The Court accepts that the memorandum, read in isolation and by its own terms, concerns visa issuance rather than the removal of noncitizens already here. CITR does not contend otherwise, having acknowledged that the May Memo, standing alone, “appear[s] to be restricted to visa

14

issuance.” Hr’g Tr. at 7:8–10. Yet that concession yields Defendants far less than they suppose. CITR has never claimed that the May Memo exhausts the policy’s scope. It instead asserts that the Department expanded and applied the announced policy through its December actions — directing consular officers to deny visas to those engaged in content-moderation work, and then deploying the same policy to “render deportable” noncitizens already here for that same protected activity. See Pl. Mot. at 6–8; ECF No. 48 (Pl. Reply) at 3–4.

The record makes the connection plain. The first link is the cable, reportedly transmitted to consular officers on December 2, which invoked the May policy and directed officers to scrutinize visa applicants for involvement in “combatting misinformation, disinformation or false narratives, fact-checking, content moderation, compliance, and trust and safety.” NPR Article at ECF p. 4. Such individuals were to be found “responsible for, or complicit in, censorship” and denied visas. Id. at ECF p. 3. The second link is the removal determination of December 19, by which the Secretary found Imran Ahmed deportable under 8 U.S.C. § 1227(a)(4)(C) on the same foreign-policy ground articulated in the May Memo. See Dec. Action Mem. at 5–6.

That the removal side of this course of conduct belongs to the same policy as the visa side is established, in the end, by the Government’s own words. A deportability determination under § 1227(a)(4)(C) must rest on the Secretary’s determination that the noncitizen’s presence or activities would have potentially serious adverse foreign-policy consequences for the United States. See supra Section I.A. For Imran Ahmed’s deportation, the Secretary grounded that finding in the May policy. The December Action Memo stated that the deportability determination was “[c]onsistent with those foreign policy objectives articulated in the May 2025 3C policy.” Dec. Action Mem. at 5. In a subsequent memorandum to then-Secretary Noem recommending Ahmed’s removal, Secretary Rubio explained that the determination was “based

15

on the same foreign policy objectives articulated in [the] May announcement of a policy to restrict visa issuance under INA section 212(a)(3)(C).” Id. at 7. Before Congress, Under Secretary Rogers identified Ahmed as “one of the targets of our recent 3C sanctions policy.” ECF No. 48-1 (Supplemental Declaration of Carrie DeCell), ¶ 6. And a December 23 series of public posts announced inadmissibility findings, visa restrictions, and deportability determinations together as enforcement of a single campaign against censorship. See Enforcement Action Press Release; ECF Nos. 11-25–30 (Rubio & Rogers X Posts). These are not the descriptions of an enforcement action that stands apart from the policy; they are the descriptions of an enforcement action that carries the policy forward.

Pressed at oral argument to reconcile this language with the severance it urges, Defendants’ counsel could say only that although the determination is “based on the same objectives,” it was nonetheless “not based on the policy itself.” Hr’g Tr. at 23:9–15. That is a distinction without a difference. A shared objective, without more, might not establish a unified policy. Here, however, the responsible officials invoked the May policy by name, applied its central substantive criteria, identified Ahmed as one of its targets, and announced the visa and removal actions together. Defendants offer no persuasive explanation for those repeated connections if the Ahmed determination was, as they now maintain, wholly freestanding.

The Government resists this conclusion principally on the ground that the Ahmed deportation was made on an “individualized” basis under the Secretary’s discretionary authority. See Opp. at 5, 25. In their telling, that determination stands apart from the policy, which they describe as a measure concerned only with visa restrictions. Id. at 4–5, 25. The individualized character of the determination, however, cannot be what places removal outside the policy because the visa restrictions the Government concedes implement the policy also rest on

16

individualized determinations. The inadmissibility provision, no less than the deportability provision, conditions the exclusion of a noncitizen for protected “beliefs, statements, or associations” on the Secretary’s “personal[]” determination that the noncitizen’s admission “would compromise a compelling United States foreign policy interest.” 8 U.S.C. § 1182(a)(3)(C)(iii). The deportability provision incorporates that requirement by direct reference. Id., § 1227(a)(4)(C)(ii). Congress thus yoked the two grounds to the same standard, making an individualized, personal determination by the Secretary the predicate of exclusion and removal alike. If the need for such a determination placed the removal action outside the policy, it would equally place outside the policy the visa restrictions the Government concedes make up the policy.

At this stage, then, the challenged conduct is best understood not as a visa policy trailed by unrelated removal decisions, but as a single policy aimed at noncitizens the Department deems complicit in censorship, pursued through two mechanisms: visa denial for those seeking admission and removal of those already here. Indeed, the deportability ground borrows the inadmissibility ground’s substantive criteria by direct reference, and the Department’s own documents tie both to the same announced objective.

2. Standing

With the challenged policy thus defined, the Court turns to Defendants’ threshold objection that CITR lacks standing to bring the suit. They contend that the policy neither regulates the Coalition itself nor directly interferes with its core activities, and that any injury to CITR depends instead on the voluntary choices of its members and other noncitizens. See Opp. at 15–18. CITR, for its part, offers several routes to standing — associational through its citizen and noncitizen members, and organizational in its own right. See Pl. Mot. at 13–27; see also Ctr.

17

for Responsible Sci. v. Gottlieb, 346 F. Supp. 3d 29, 36 (D.D.C. 2018) (delineating organizational and associational standing). The Court need travel only one. Because Plaintiff has shown an injury to its own organizational activities, the Court finds standing on that ground and does not reach the associational theories. See Ctr. for Biological Diversity v. EPA, 56 F.4th 55, 69 (D.C. Cir. 2022) (declining to address organizational standing after finding associational standing).

The constitutional minimum is settled. To have standing, a plaintiff must show that she “(1) suffered an injury in fact, (2) that is fairly traceable to the challenged conduct of the defendant, and (3) that is likely to be redressed by a favorable judicial decision.” Spokeo, Inc. v. Robins, 578 U.S. 330, 338 (2016). CITR has established each.

a. Organizational Injury in Fact To establish an injury in fact of its own, an organization must satisfy the two-part test set forth by the Circuit. The Court first asks “whether the agency’s action or omission to act injured the [organization’s] interest” and then “whether the organization used its resources to counteract that harm.” PETA v. USDA, 797 F.3d 1087, 1094 (D.C. Cir. 2015) (quotation marks and citation omitted). Under the first prong, the challenged conduct must “perceptibly impair[] the organization’s ability to provide services,” Food & Water Watch, Inc. v. Vilsack, 808 F.3d 905, 919 (D.C. Cir. 2015) (quoting Turlock Irrigation Dist. v. FERC, 786 F.3d 18, 24 (D.C. Cir. 2015)), “inhibit[]” its “daily operations,” PETA, 797 F.3d at 1094 (quotation marks and citation omitted), or “ma[k]e the organization’s activities more difficult.” Nat’l Treasury Emps. Union v. United States, 101 F.3d 1423, 1430 (D.C. Cir. 1996) (emphasis omitted). The second prong is met where the organization “undertook the expenditures in response to, and to counteract, the

18

effects of the defendant’s [challenged conduct] rather than in anticipation of litigation.” Equal Rts. Ctr. v. Post Props., Inc., 633 F.3d 1136, 1140 (D.C. Cir. 2011).

CITR finds success on both fronts, and the reason starts with what the policy targets. The Coalition exists to convene, sustain, and build the field of independent technology researchers — those who analyze public data, report on disinformation, and press technology platforms over content moderation to mitigate user harm. See Geurkink Decl., ¶¶ 4–9. The policy renders the people who do that work subject to visa revocation, exclusion, and deportation, and it has already been deployed against the leaders of two CITR organizational members. See Dec. Action Mem. at 5–6; Compl., ¶¶ 57–58. An organization whose purpose is to hold such a community together does not stand at a remove from a policy that takes the community as its object; the impairment reaches the substance of what CITR does. The record bears that out across its operations.

Begin with CITR’s reporting and public-facing research. Since the policy’s enforcement, noncitizen members have declined to attach their names to CITR’s own State of Independent Tech Research Report, and the Coalition has been unable to find members willing to speak publicly either about their work or the effects of the policy on independent technology research. See Geurkink Decl., ¶ 28. That is not merely a diminished appetite for advocacy. It is the degradation of a concrete organizational work product: CITR cannot research, publish, and publicize as its operations require when the contributors that such work depends upon go silent.

CITR’s convening function has fared no better. It ordinarily holds one or two events a month and an annual summit drawing its global membership, the means by which it gives researchers room to collaborate and to confront the threats they share. Id., ¶ 8. The policy has hollowed that function out. U.S.-based noncitizen members declined to attend the 2025 summit in Berlin for fear of being denied reentry, thinning participation and hampering CITR’s ability to

19

coordinate work across its membership. Id., ¶ 19. CITR has suspended its U.S.-based events altogether, unable to hold the inclusive, cross-border gatherings that are their whole purpose while the policy deters attendance. Id., ¶¶ 16–18. And since the December enforcement, members have grown unwilling to take part candidly even in the events that go forward: many no longer sign in, no longer document their work as a group, and decline to share what they know with anyone not in attendance. Id., ¶ 21.

Members have withdrawn as well from public association with CITR itself. Several —

among them an individual identified as Member C — have asked that their names come off CITR’s website and have stopped attending meetings, telling Coalition leadership that association no longer feels safe once leaders of member organizations have themselves become targets. Id., ¶ 24. That severs the very tie between organization and member that CITR exists to maintain, and it reflects “a direct conflict between [Defendants’] conduct and the organization’s mission.” Elec. Priv. Info. Ctr. v. FAA, 892 F.3d 1249, 1255 (D.C. Cir. 2018) (emphasis omitted) (quoting Nat’l Treasury Emps. Union, 101 F.3d at 1430).

Defendants reduce this all to a dip in event attendance — no more than “ipse dixit anxiety.” Opp. at 16–17 (citing Already, LLC v. Nike, Inc., 568 U.S. 85, 97 (2013)). Yet that framing reduces a broad operational injury to a triviality it does not resemble. The participation CITR has lost — contributions to its reports, voices to the press, public association under its banner, candor at its convenings — is the participation on which its reporting and convening depend, and its loss is an injury to what the organization does, not a complaint about what its members would prefer.

Nor does that withdrawal rest on unfounded fear. Defendants insist that CITR cannot show that its members are targeted, making any chill “subjective” and insufficient under Clapper

20

v. Amnesty International USA, 568 U.S. 398 (2013). See Opp. at 13–14, 17–18. That, however, ignores that the policy has already been enforced against the leaders of two CITR member organizations, and Defendants have announced that the Department “stands ready and willing to expand” its list of targets. See ECF No. 11-28 (Melford Sanction X Post); Dec. Action Mem. at 5–6; Enforcement Action Press Release at ECF p. 2. In other words, members who do the same work as those already sanctioned need not speculate when they face the same fate, and their withdrawal from the Coalition’s work is the foreseeable result. Having shown both that its operations are impaired and that the threat driving that impairment is real and not conjectural, CITR clears the first prong.

The second is likewise satisfied, as Plaintiff has expended resources to counteract these harms and preserve operations. To keep its members together as best it can, CITR has split its annual summit into two parallel gatherings — one abroad and one in the United States — so that no members need travel internationally and place themselves at risk. See Geurkink Decl., ¶ 20. That arrangement creates two separate cost structures and will charge the organization tens of thousands of dollars that otherwise would have gone toward staff positions it can no longer afford. Id. The Coalition has also had to spend staff time making participation possible again — building ad hoc safety processes for public-facing work, counseling members on the policy’s implications, and convening noncitizen researchers to address its effects. Id., ¶¶ 29–32. These were “not part of [CITR’s] normal annual expenditures,” and they have worked “a consequent drain on the organization’s resources.” Am. Anti-Vivisection Soc’y v. USDA, 946 F.3d 615, 619 (D.C. Cir. 2020) (citation omitted). The diversion has come at a cost that can be named: CITR has shelved the capacity-building workshops on data access it meant to launch and has had no time to pursue public education on the funding cuts facing its field. See Geurkink Decl., ¶ 33.

21

These are not litigation costs or freestanding advocacy expenses, but diversions from existing programs to counteract ongoing conduct — the very paradigm the second prong contemplates. See Humane Soc’y of U.S. v. USDA, 41 F.4th 564, 567–68 (D.C. Cir. 2022).

Defendants answer that these are advocacy expenditures, which cannot ground standing, invoking the rule that an organization may not manufacture injury by spending to oppose a policy it dislikes. See Opp. at 17–18 (citing Turlock, 786 F.3d at 24; Ctr. for L. & Educ. v. Dep’t of Educ., 396 F.3d 1152, 1162 n.4 (D.C. Cir. 2005); and Food & Drug Admin. v. All. for Hippocratic Med., 602 U.S. 367, 394 (2024)). The rule is sound but does not apply to this case. Turlock and Center for Law & Education, like Alliance for Hippocratic Medicine, concern organizations that disagree with government action, run up the costs of opposing it, and then offer those costs as injury. The plaintiffs in Alliance for Hippocratic Medicine, for example, had suffered no concrete disruption to their own operations; their costs — studies, citizen petitions, and public education — were efforts to oppose FDA’s action, not to repair operations the agency had disrupted. See 602 U.S. at 394–95.

CITR’s outlays are of a different order. They are not the price of opposing the policy in this litigation but the price of keeping impaired operations running: holding two events where one once served, building processes so members will contribute to existing programs, and working around the participation the policy has chilled. That an organization “voluntarily, or willfully, diverts its resources” in that way “does not automatically mean that it cannot suffer an injury sufficient to confer standing.” Equal Rts. Ctr., 633 F.3d at 1140 (cleaned up). What matters is that the expenditures counteract a direct impairment rather than advocate against a disfavored policy.

22

b. Causation

As for causation, Defendants argue that CITR’s injuries stem from its members’

voluntary choices rather than from the policy itself and are therefore not fairly traceable to the Government’s conduct. See Opp. at 15–16. The point has some force, but not enough. That CITR’s injuries reach it through members’ responses does not sever their connection to the policy. An injury that follows from the predictable reaction of others to government action is fairly traceable to that action. See Dep’t of Com. v. New York, 588 U.S. 752, 768 (2019); see also Nat’l Wrestling Coaches Ass’n v. Dep’t of Educ., 366 F.3d 930, 941 (D.C. Cir. 2004) (recognizing that plaintiffs not directly regulated may establish standing by showing causal relationship between government policy and third-party conduct), abrogation on other grounds recognized by Perry Cap. LLC v. Mnuchin, 864 F.3d 591 (D.C. Cir. 2017); N.Y. Republican State Comm. v. SEC, 927 F.3d 499, 504–05 (D.C. Cir. 2019).

The reaction here was not merely predictable; it was all but ordered. Announcing the enforcement actions against leaders of two CITR member organizations, Rubio warned others engaged in the same work to “reverse course” or face the same. See Enforcement Action Press Release at ECF p. 2. The record shows that the message landed. Member A has refrained from international travel, including to CITR’s 2025 summit in Berlin, because of fear of being denied reentry under the policy; has limited public advocacy with CITR to a behind-the-scenes role because of fear of detention and deportation; and says that he or she would be substantially more likely to resume public association with CITR were the policy no longer in place. See ECF No. 11-5 (Declaration of CITR Member A), ¶¶ 9–11. Dr. Emma L. Briant, a U.K. citizen and Visiting Associate Professor at Notre Dame, likewise avers that the policy has caused her to self- censor in her writing and public speaking, hesitate to travel internationally, and evaluate even

23

domestic speaking invitations against the risk of detention or deportation. See ECF No. 11-6 (Declaration of Emma L. Briant), ¶¶ 3, 5, 8–10. Those declarations confirm CITR’s own account: members have stepped back from CITR’s reporting, events, and public-facing work because they fear becoming the next targets. See Geurkink Decl., ¶¶ 19, 21, 24, 28.

c. Redressability Redressability follows, and Defendants scarcely contest it. Having trained their fire on injury and causation, they offer no developed argument that a favorable decision would leave CITR’s harms unredressed. The point is clear in any event: “[I]f a government action causes an injury, enjoining the action usually will redress that injury.” Carpenters Indus. Council v. Zinke, 854 F.3d 1, 6 n.1 (D.C. Cir. 2017). A stay of the policy would temporarily lift the threat that has driven CITR’s members from its reports, its events, and its public-education efforts, and the record reflects that they would return to those activities were the policy no longer in force. See Geurkink Decl., ¶¶ 18, 22–24. CITR’s injury would thus be redressed by the relief it seeks.

3. Jurisdiction

That CITR has standing to sue does not end the threshold inquiries, for the Government’s next line of defense is jurisdictional. It marshals four provisions of the INA: three that channel review of individual visa and removal decisions away from district courts — 8 U.S.C. § 1252(g), § 1252(b)(9) (with its companion, § 1252(a)(5)), and § 1201(i) — and a fourth, § 1252(f)(1), said to limit the relief the Court may order against the removal statutes’ operation even if jurisdiction lies. See Opp. at 32–39. The four share a premise: that this suit is, in substance, a challenge to the Government’s individual visa and removal decisions of the kind Congress routed into the petition-for-review process and insulated from broad injunctive relief.

24

Before addressing each provision, the Court notes that two features of this case answer the Government’s four arguments. First, CITR does not ask the Court to review any particular visa revocation, removal proceeding, order of removal, or decision to seek a specific noncitizen’s removal. It instead challenges the lawfulness of the policy itself, and it does so as an organization that holds no visa, is not subject to removal, and whose members (the Government does not dispute) are not presently in removal proceedings. See Opp. at 7. Sections 1252(g), 1252(b)(9), and 1201(i) are each aimed at the judicial review of an individual noncitizen’s visa or removal, and a challenge to a Government-wide policy — brought by a party against which the enforcement machinery will never engage — is not that. The second feature answers § 1252(f)(1), which speaks not to what a court may hear but to what it may order. Whatever the reach of that provision over the injunction CITR requests, it does not touch the stay under 5 U.S.C. § 705, which would redress the Coalition’s injury.

With those two overarching points in mind, the Court takes the three channeling provisions first and then turns to the question of remedy.

a. Section 1252(g)

The Government leads with § 1252(g). By its terms, this section withdraws court jurisdiction from “any cause or claim by or on behalf of any alien arising from the decision or action by the Attorney General to commence proceedings, adjudicate cases, or execute removal orders.” 8 U.S.C. § 1252(g). Because the policy may culminate in removal for some noncitizens it targets, the Government contends that a suit challenging the policy is a claim “arising from” the decision to commence proceedings, and so one the statute bars.

The Supreme Court has read § 1252(g) to do far less work than the Government supposes. In Reno v. American-Arab Anti-Discrimination Committee (AADC), 525 U.S. 471

25

(1999), the Court explained that the provision reaches “only . . . three discrete actions”: decisions to “commence proceedings, adjudicate cases, or execute removal orders.” Id. at 482. It does not extend to “all claims arising from deportation proceedings,” and the Court has rejected as “implausible” any reading that would give the provision that “general jurisdictional” sweep. Id. The bar protects the exercise of prosecutorial discretion at three enumerated points. It does not immunize from review every antecedent policy choice that may eventually contribute to a removal.

The Government’s contrary argument overlooks that § 1252(g) is limited twice over, both in what it covers and in whose claims it channels. The provision reaches only the three discrete actions, and only when the challenge comes from or on behalf of the noncitizen those actions target. AADC illustrates the point. The respondents there were noncitizens whom the Government had placed in deportation proceedings and who sued to halt those proceedings, contending that they had been singled out for their affiliation with a politically disfavored group in violation of the First Amendment. See AADC, 525 U.S. at 473–74. The Court held that their challenge fell “squarely within § 1252(g)” because it attacked the “decision to ‘commence proceedings’” against them, and it so held even though the claim sounded in the First Amendment and even though the respondents urged that deferring review would chill their protected speech. Id. at 487–88. What brought the suit within the bar was its object — an effort by noncitizens to arrest the commencement of their own removal — not the constitutional theory on which it rested.

CITR’s suit does not fit that paradigm. The Coalition is not a noncitizen in or facing removal proceedings; it does not ask the Court to stop the Government from commencing, adjudicating, or executing anyone’s removal; and the relief it seeks runs to the policy itself, not

26

to any removal proceeding. Where the AADC respondents stood inside the removal process and sought to halt it, CITR stands outside that process altogether. It is, in the words of a court that confronted a materially similar challenge, not an “alien[] seeking to undo or prevent removal proceedings commenced against” it, and the narrow bar of § 1252(g) “does not apply to constitutional challenges brought by one who is not the alien subject to the three discrete decisions . . . or one who is not bringing a challenge to such actions on the alien’s behalf.” NWDC Resistance v. ICE, 493 F. Supp. 3d 1003, 1011 (W.D. Wash. 2020); see also Am. Ass’n of Univ. Professors v. Rubio, 780 F. Supp. 3d 350, 373 (D. Mass. 2025) (Section 1252(g) did not bar an organizational challenge to an ideological-deportation policy). That CITR’s claim invokes the First Amendment is immaterial, as § 1252(g) attaches to the three enumerated actions, not to the theory of the claim. The provision channels challenges arising from those actions when brought by or on behalf of the noncitizens whom the removal machinery is processing, and CITR is not among them.

This reading fits the purpose the bar serves. Section 1252(g) shields the kind of classic prosecutorial decisions that are “particularly ill-suited to judicial review” because they turn on judgments about “the strength of the case, the prosecution’s general deterrence value, the Government’s enforcement priorities, and the case’s relationship to the Government’s overall enforcement plan.” AADC, 525 U.S. at 490 (quoting Wayte v. United States, 470 U.S. 598, 607 (1985)); see also Escobar Molina v. DHS, 811 F. Supp. 3d 1, 41 (D.D.C. 2025) (drawing this distinction). CITR asks for nothing of the sort. It does not seek to unwind any individual charging decision, and its claims turn on legal questions — namely, whether the policy comports with the First Amendment and the Administrative Procedure Act — that lie “well within courts’ normal purview.” Escobar Molina, 811 F. Supp. 3d at 41.

27

The decisions the Government points to only prove the rule. In Tazu v. Attorney General, 975 F.3d 292 (3d Cir. 2020), the petitioner had a final order of removal and was arrested so that the order could at last be executed once the Government had obtained a travel document. The Third Circuit held that § 1252(g) barred his challenge precisely because “[]detaining Tazu was simply the enforcement mechanism the Attorney General picked to execute his removal.” Id. at 298–99. The remaining authorities are of a piece. See, e.g., Hamama v. Adducci, 912 F.3d 869, 874–76 (6th Cir. 2018) (Section 1252(g) barred district-court relief where noncitizens challenged Government’s “decision to execute final orders”); Rauda v. Jennings, 55 F.4th 773, 777–80 (9th Cir. 2022) (same where noncitizen challenged discretionary decision to execute his removal order rather than await further administrative proceedings); Camarena v. Director, ICE, 988 F.3d 1268, 1270–74 (11th Cir. 2021) (same where noncitizens sought to halt execution of valid removal orders while pursuing provisional unlawful-presence waivers). In each, a noncitizen sought district-court intervention in an enforcement step directed at his own removal.

Because CITR challenges a policy rather than any decision to commence, adjudicate, or execute removal directed at one of its members, § 1252(g) does not block the way.

b. Sections 1252(a)(5) and 1252(b)(9)

The Government next turns to § 1252(b)(9) and its companion, § 1252(a)(5), the provisions that channel challenges to removal into a single petition for review in the court of appeals. Because the policy may one day be invoked in removal proceedings, the Government argues, CITR’s claims “aris[e] from” an action taken to remove and belong there rather than here. See Opp. at 36–38. This too fails to raise the jurisdictional walls.

28

The structure of these provisions, beginning with their titles, shows their limited reach.

Section 1252 is captioned “Judicial review of orders of removal,” and each relevant subsection returns to that subject. Subsection (a)(5), “Exclusive means of review,” provides that “a petition for review filed with an appropriate court of appeals . . . shall be the sole and exclusive means for judicial review of an order of removal.” Subsection (b), “Requirements for review of orders of removal,” applies “[w]ith respect to review of an order of removal under subsection (a)(1),” and its paragraph (9), “Consolidation of questions for judicial review,” provides that “[j]udicial review of all questions of law and fact . . . arising from any action taken or proceeding brought to remove an alien . . . shall be available only in judicial review of a final order.” Read together, the provisions route to the courts of appeals the legal and factual questions bound up in an individual’s removal, so that “district courts do not become involved in reviewing” the records that the immigration courts develop. Escobar Molina, 811 F. Supp. 3d at 38.

Jennings v. Rodriguez, 583 U.S. 281 (2018), confirms that § 1252(b)(9) does not sweep in every claim that has some connection to removal. The plaintiffs there were detained noncitizens who challenged the Government’s authority to detain them for prolonged periods without bond hearings. Id. at 286 (plurality opinion). They did not seek review of any removal order or of the process for determining their removability, raising the question of whether § 1252(b)(9) nonetheless barred their claims. Id. at 292–93.

Writing for a three-Justice plurality, Justice Alito answered that it did not, warning that an “expansive interpretation of § 1252(b)(9)” would yield “staggering” results, reaching claims as far afield as challenges to “inhumane conditions of confinement.” Id. at 293. It was enough, the plurality explained, that the respondents were “not asking for review of an order of removal,” were “not challenging the decision . . . to seek removal,” and were “not even challenging any

29

part of the process by which their removability will be determined.” Id. at 294. Justice Breyer’s opinion for the three dissenting Justices, though disagreeing on the merits, took the same view of § 1252(b)(9), concluding that the provision did not apply “by its terms” because the respondents had challenged “their detention without bail, not an order of removal.” Id. at 355 (Breyer, J., dissenting). Six Justices thus agreed that the provision does not bar claims that stand apart from the removal process itself. The Court has since reaffirmed that § 1252(b)(9) “does not present a jurisdictional bar” where those bringing suit “are not asking for review of an order of removal, the decision . . . to seek removal, or the process by which . . . removability will be determined.” Dep’t of Homeland Sec. v. Regents of the Univ. of Cal., 591 U.S. 1, 19 (2020) (quotation marks omitted) (quoting Jennings, 583 U.S. at 294–95).

Courts in this district have applied the same principles to challenges akin to CITR’s. For instance, in O.A. v. Trump, 404 F. Supp. 3d 109 (D.D.C. 2019), the plaintiffs did not seek review of any removal order but instead mounted “a facial challenge to the validity of a regulation of general applicability.” Id. at 128. Because § 1252(a)(5) concerns the means to obtain “judicial review of an order of removal,” it had no application to that kind of challenge. Id. at 128–29. Section 1252(b)(9) fared no better. It would be “implausible,” the court reasoned, “to construe § 1252(b)(9)’s mention of an ‘action taken’ or a ‘proceeding brought to remove an alien’ as shorthand for any agency action that might ultimately facilitate the removal of a class of aliens.” Id. at 132–33; see also Escobar Molina, 811 F. Supp. 3d at 38–41 (similarly finding § 1252(b)(9) does not bar policy challenges). The same is true here. CITR challenges an agency policy, not any action taken to remove a particular member. The prospect that the policy may later contribute to some noncitizen’s removal does not transform a challenge to the policy into the review of an order of removal.

30

The Government’s reliance on Khalil v. President, United States of America, 164 F.4th 259 (3d Cir. 2026), does not change the analysis. See Opp. at 38. To be sure, the Third Circuit read § 1252(b)(9) broadly, holding that legal questions must await petition-for-review proceedings when they can be meaningfully reviewed there, even if the plaintiff alleges present injuries that cannot later be remedied. See Khalil, 164 F.4th at 274–76. Yet Khalil was himself arrested, detained, charged as removable, and placed in removal proceedings, and the court concluded that the legal questions underlying his claims could be examined in a petition for review of any final order. Id. at 266–68, 273–77. CITR is not in removal proceedings, cannot file a petition for review, and cannot obtain review of its organizational claims in a proceeding to which it is not a party. The chill the Coalition alleges, furthermore, works its harm precisely because its members forgo activity that might place them in removal proceedings. Without that activity, no proceeding will ever arise in which the legal questions could be raised. Sections 1252(a)(5) and 1252(b)(9) accordingly do not strip the Court of jurisdiction.

c. Section 1201(i)

For largely the same reasons, the Government gets no further with § 1201(i). That provision forecloses judicial review “of a revocation” of a visa, “except in the context of a removal proceeding if such revocation provides the sole ground for removal.” 8 U.S.C. § 1201(i). Because the policy operates in part through visa revocations, Defendants urge that CITR’s challenge to the policy is, in substance, a forbidden attack on those revocations. See Opp. at 36–37. It is not. Section 1201(i) speaks to review of a particular visa revocation, and it channels that challenge into the removal proceeding in which the revocation is put to use. It does not speak to pre-enforcement review of a policy that may later produce revocations in individual cases.

31

CITR does not ask the Court to review or undo any particular visa revocation. It challenges the Government’s policy in toto, and the relief it seeks runs to the policy itself. As with § 1252(b)(9), the possibility that the policy may be applied through downstream immigration decisions does not convert a challenge to the policy into review of those decisions. See O.A., 404 F. Supp. 3d at 132–33 (distinguishing challenge to rulemaking of general applicability from challenge to “action taken” or “proceeding brought” to remove particular noncitizen); cf. AADC, 525 U.S. at 482 (declining to treat antecedent steps in deportation process as discrete actions Congress placed beyond district-court review). Because CITR seeks no review of any visa revocation, § 1201(i) does not deprive the Court of jurisdiction.

d. Section 1252(f)(1)

The Government’s final § 1252 argument is of a different stripe. Section 1252(f)(1) does not speak to the Court’s jurisdiction to hear CITR’s claims but limits the relief a court may grant. The provision states that, except for the Supreme Court, “no court . . . shall have jurisdiction or authority to enjoin or restrain the operation of” the covered removal provisions — §§ 1221 through 1232 — “other than with respect to the application of such provisions to an individual alien against whom proceedings . . . have been initiated.” Invoking that provision, Defendants maintain that the Court cannot restrain their use of § 1227(a)(4)(C) or the removal procedures of § 1229a, whether through a preliminary injunction or a § 705 stay. See Opp. at 38–39. But whatever force § 1252(f)(1) may have as to coercive injunctive relief, it does not bar the § 705 stay CITR seeks here.

Just last month, our Circuit so held in Make the Road New York v. Mullin, --- F.4th ---, 2026 WL 1792978 (D.C. Cir. June 23, 2026). The Government there asserted that § 1252(f)(1) deprives lower courts of authority to issue a § 705 stay, and the Court of Appeals rejected that

32

argument. Id. at *10–13. Section 1252(f)(1), the court explained, bars orders that “enjoin or restrain” the operation of the covered provisions, and those terms refer to formally coercive relief — judicial commands that direct or prohibit a party’s conduct. Id. at *11–12. A stay is different. It “does not order parties to do anything”; instead, it operates on the legal status of the challenged agency action by temporarily suspending its operative force. Id. at *11; see also Nken, 556 U.S. at 428–29 (differentiating “enjoin” from “stay”). On that basis, the court held that § 1252(f)(1) “bars only judicial commands that coerce a party’s behavior — orders that ‘enjoin or restrain,’ not orders that ‘stay.’” Make the Rd. N.Y., 2026 WL 1792978, at *13.

The Court recognizes that, from CITR’s vantage, the two remedies converge: whether enjoined or stayed, the policy will not operate while this case proceeds. But Congress drew the line in § 1252(f)(1) by remedy, not by result, withdrawing only the authority to “enjoin or restrain” officials from administering the covered provisions. A stay of the policy does neither. It commands no official and — because it suspends an agency policy rather than any statute — leaves §§ 1221 through 1232 fully operative. That is enough for present purposes. CITR seeks, at a minimum, a § 705 stay of the policy pending review. Under our Circuit’s recent precedent, § 1252(f)(1) does not deprive the Court of authority to grant that relief. It therefore need not decide at this stage whether, or to what extent, § 1252(f)(1) would limit any additional coercive injunction directed at the operation of §§ 1227 or 1229a.

* * *

As a result, the Court concludes that CITR has established that this Court has jurisdiction over its claims and authority to grant the interim relief necessary to preserve the status quo pending review. On now to the merits.

33

4. First Amendment

A word on the order of analysis is warranted. Because the relief CITR seeks — a stay under 5 U.S.C. § 705 — is a creature of the APA, one might expect that statute to lead the analysis. Here, however, the APA is the vehicle rather than the destination.

Plaintiff’s APA count presses several theories: that the policy is arbitrary and capricious, see 5 U.S.C. § 706(2)(A); in excess of statutory authority, id., § 706(2)(C); and contrary to constitutional right. Id., § 706(2)(B); see also Pl. Mot. at 38–44. That last theory does not stand apart from CITR’s First Amendment claim; instead, it creates the cause of action for Plaintiff to challenge the purported violation of its First Amendment rights. See Make the Rd. N.Y., 2026 WL 1792978, at *9 (“Constitutional challenges to agency action are . . . resolved within the APA’s judicial-review framework.”); Trudeau v. FTC, 456 F.3d 178, 188–90 (D.C. Cir. 2006) (APA furnishes “more inclusive” and “more expansive” vehicle for constitutional challenges to agency action). As the constitutional question forms the thrust of this dispute, the Court begins with the First Amendment, returning afterward to the APA-specific requirements that condition review under § 705.

CITR’s First Amendment challenge advances two theories. First, it says that the policy unlawfully discriminates against disfavored viewpoints about online content. See Pl. Mot. at 30– 32. Second, the Coalition believes that Defendants’ threats of visa denial, exclusion, and removal amount to an unconstitutional campaign of coercion. Id. at 34–35. The Government responds that the policy does no such thing. In its telling, the policy targets conduct, not speech; foreign censorship, not content-moderation research; and immigration decisions that receive substantial deference, not domestic regulation of protected expression. See Opp. at 20–25.

34

As explained below, the record supports CITR’s first theory: the policy is likely viewpoint discriminatory in a substantial share of its applications. As expanded in December and applied since, it does not stop at individuals who use sovereign authority to suppress protected expression in the United States. It reaches private researchers, advocates, nonprofit leaders, and trust-and-safety professionals whose asserted “censorship” consists of studying platforms, reporting on disinformation, petitioning for access to platform data, and urging platforms or advertisers to act on what they find. Because CITR is likely to show that this operation discriminates on the basis of viewpoint, the Court need not decide whether the same course of conduct independently amounts to unconstitutional coercion.

a. Protected Interests and Facial Review The First Amendment reflects “a profound national commitment to the principle that debate on public issues should be uninhibited, robust, and wide-open.” N.Y. Times Co. v. Sullivan, 376 U.S. 254, 270 (1964). That commitment is not confined to stump speeches, editorials, or familiar forms of political advocacy. It protects the gathering and publication of information, Branzburg v. Hayes, 408 U.S. 665, 681 (1972); Bartnicki v. Vopper, 532 U.S. 514, 527–28 (2001), the petitioning of officials, BE&K Constr. Co. v. NLRB, 536 U.S. 516, 524–25 (2002), and the right to associate with others in pursuit of shared political, social, and educational ends. NAACP v. Alabama ex rel. Patterson, 357 U.S. 449, 460–63 (1958); Roberts v. U.S. Jaycees, 468 U.S. 609, 622 (1984). Nor do those protections fade with the advent of new technologies. “‘[W]hatever the challenges of applying the Constitution to ever-advancing technology, the basic principles’ of the First Amendment ‘do not vary.’” Moody v. NetChoice, LLC, 603 U.S. 707, 741 (2024) (quoting Brown v. Ent. Merchs. Ass’n, 564 U.S. 786, 790 (2011)).

35

Those principles cover the activity chilled here. CITR’s work depends on researchers who study how platforms structure public debate, report on misinformation and disinformation, advocate for access to platform data, petition officials, speak to the press, and collaborate with one another to set standards and press for reform. Some of that work culminates in reports, interviews, comments, petitions, and testimony. Some of it occurs before publication, in the candid exchange among researchers and organizations that makes public-facing work possible. Those activities, at least as reflected in this record, fall within the Amendment’s protection for speech, publication, petitioning, and expressive association. They also sit directly within the contested public debate over how online platforms structure discourse and whether, when, and how they should moderate harmful or false content. See, e.g., Media Matters for Am. v. Paxton, 138 F.4th 563, 584 (D.C. Cir. 2025) (treating reporting on “alleged political extremism on a popular social media platform” as “quintessential First Amendment activit[y]”).

CITR’s asserted injury is therefore not merely derivative of what its noncitizen members might say or what CITR might hear. The policy allegedly impairs CITR’s own work: who will contribute to its reports, what those reports can say, who will attach their names to them, and whether researchers will participate in the convenings and candid exchanges from which CITR’s public work emerges. See supra Section III.A.2.a. That is itself a First Amendment burden, as the Amendment protects both an organization’s creation and dissemination of information, Sorrell v. IMS Health Inc., 564 U.S. 552, 570 (2011), and the associational activity that makes collective speech possible. NAACP, 357 U.S. at 460 (“Effective advocacy of both public and private points of view, particularly controversial ones, is undeniably enhanced by group association, as this Court has more than once recognized by remarking upon the close nexus between the freedoms of speech and assembly.”).

36

A second principle is equally settled. The Government may not burden protected speech because it disfavors one side of a public debate. Viewpoint discrimination is “an egregious form of content discrimination,” Rosenberger v. Rector & Visitors of Univ. of Va., 515 U.S. 819, 829 (1995), and the First Amendment forbids “the [G]overnment from ‘tilt[ing] public debate in a preferred direction.’” Moody, 603 U.S. at 741 (quoting Sorrell, 564 U.S. at 578–79). The point is not simply that officials may not silence ideas they dislike. It is that they may not use the tools of government to remake the private-speech environment according to their preferred balance of expression. “On the spectrum of dangers to free expression,” the Supreme Court has warned, “there are few greater than allowing the government to change the speech of private actors in order to achieve its own conception of speech nirvana.” Id. at 741–42.

Because CITR brings a facial challenge, those principles must be applied through the discipline facial review requires. Moody supplies the path. A court must first identify the challenged measure’s scope: “[w]hat activities, by what actors” it reaches. Id. at 724. It must then decide which applications violate the First Amendment and “measure them against the rest.” Id. at 725. CITR succeeds only if “a substantial number of the policy’s applications are unconstitutional, judged in relation to the policy’s plainly legitimate sweep.” Id. at 723 (cleaned up). That test is demanding, as facial challenges are, and the burden of satisfying it belongs to CITR, which must carry it “from the text of [the policy] and from actual fact” rather than from supposition. Virginia v. Hicks, 539 U.S. 113, 122 (2003) (quoting N.Y. State Club Ass’n, Inc. v. City of New York, 487 U.S. 1, 14 (1988)). The same discipline binds the Court, which may not “speculate about ‘hypothetical’ or ‘imaginary’ cases” in comparing either side of the ledger. Wash. State Grange v. Wash. State Republican Party, 552 U.S. 442, 450 (2008). Demanding as

37

it may be, the First Amendment’s overbreadth doctrine nevertheless exists to “provide[] breathing room for free expression.” United States v. Hansen, 599 U.S. 762, 769 (2023).

b. The Policy’s Scope The first question asks what this policy reaches. The Government describes a measure aimed at those who facilitate foreign-government censorship of Americans’ speech. See Opp. at 20–22; Hr’g Tr. at 28:22–24. CITR, conversely, depicts one that extends to private researchers, advocates, and platform professionals whose work merely favors content moderation. See Pl. Mot. at 4–9, 10–13. The choice between those accounts cannot be made from the May Memo alone, for the challenged measure is that memorandum as expanded in December and applied since. See supra Section III.A.1. The May Memo set the policy’s terms, authorizing action against those “responsible for, or complicit in,” censorship of “protected expression.” May Mem. at ECF p. 1. The inquiry thus turns on the meaning those terms have taken on in the Department’s hands.

The December cable supplies part of the answer. It directed consular officers to “thoroughly explore” visa applicants’ work histories, resumes, social-media profiles, and media appearances for involvement in “combatting misinformation, disinformation or false narratives, fact-checking, content moderation, compliance, and trust and safety,” and, on locating it, to pursue a finding of ineligibility. See NPR Article at ECF p. 4. Those categories do not describe the exercise of foreign sovereign power. They describe the ordinary work of researchers, fact checkers, platform employees, compliance officers, and nonprofit advocates who study, criticize, participate in, or press for content moderation. A cable that treats that work as evidence of immigration ineligibility reaches far beyond the coercive acts described in the May Memo:

38

threats of arrest, payment freezes, legal compulsion, detention, fines, and demands for private data directed at American platforms or persons in the United States. See May Mem. at ECF p. 2.

The December enforcement actions provide the rest of the picture. The Court does not review the lawfulness of any individual visa denial or deportability determination, and it decides nothing about whether any named person may ultimately be excluded or removed. The actions matter because State held them out as examples of the policy at work. Its public rationales identify the activity it treats as “complicity” in “censorship”: a report on hate speech and disinformation, advocacy directed at advertisers and platforms, disinformation-risk ratings, a petition for researcher access to platform data, a broadcast interview, and nonprofit leadership in organizations that help targets of online abuse seek removal of content aimed at them. See Compl., ¶¶ 56–60. Some of those justifications are tied to familiar First Amendment activity: reporting, speaking, petitioning, advocating for platform regulation, and associating through nonprofit leadership. Id. At least as to the private researchers and nonprofit leaders in CITR’s field, the public explanations do not identify any exercise of foreign sovereign power akin to the coercive acts the May Memo enumerates.

The concern is not limited to the five named individuals. If disinformation-risk ratings, reports on hate speech, petitions for platform-data access, advocacy, or nonprofit work seeking to limit abusive content can count as “complicity” in “censorship,” the policy has no clear stopping point short of the field itself — a concern sharpened by the Department’s announcement that it “stands ready and willing to expand” the list. See Enforcement Action Press Release at ECF p. 2. A lawful permanent resident working on a platform’s trust-and-safety team, a noncitizen researcher urging stronger disinformation labels, a compliance employee helping apply moderation rules, or an advocacy leader pressing advertisers away from sites that spread

39

falsehoods could reasonably understand the policy to place their immigration status at risk — not because they wield foreign sovereign power or facilitate its censorship, but simply because they work in content moderation.

Defendants resist this account by describing a narrower policy than the record reveals.

They maintain that the policy targets “assisting or facilitating foreign government censorship of free speech.” Hr’g Tr. at 28:23–25. That limiting principle would matter a great deal if it were the policy’s real boundary. Counsel framed it as the dividing line at argument, explaining that it has to be “some involvement with a foreign government and their censorship regimes,” id. at 28:7–9, and affirming that a person unaffiliated with any foreign government who simply “advocat[es] for content moderation or a ban on misinformation and disinformation has nothing to fear from this policy.” Id. at 30:18–21.

The trouble is that the enforcement record does not honor that line. The Government has tied none of the private researchers and nonprofit leaders targeted in December to any exercise of foreign sovereign power. Pressed on that gap, counsel did not supply the missing connection. The Government stepped back from the five examples, explaining that it lacked “the full factual records or the reasons for those determinations” and that it would not be “fair to rely on those five” in gauging the policy’s scope. Id. at 31:10–16. But Defendants cannot publicly announce examples of the policy at work, warn that the Department stands ready to expand them, and then — when those examples prove inconvenient — deny that they reveal anything about the policy’s reach. A limiting principle that the Government cannot reconcile with its own enforcement record is no limit at all.

40

c. Viewpoint Discrimination Once the policy’s scope is fixed, the relevant applications — and their defects — come into focus. The unconstitutional applications are those in which Defendants treat a noncitizen’s research, reporting, advocacy, or platform-governance work favoring greater content moderation as the basis for visa denial, exclusion, or removal. Each such application burdens the U.S. coalitions, universities, and nonprofits whose protected research, publishing, and association depend on the people the policy targets. The policy, at its core, does not burden all speech about platforms, all research into content moderation, or all advocacy about online harms. It presses its enforcement thumb against one side of the scale: the view that platforms should do more to moderate content, label disinformation, restrict abuse, share data with researchers, or take responsibility for the harms their systems amplify. The Government, in other words, has not set itself against everyone who speaks about platform governance. It has set itself against those whose work favors more moderation rather than less. A noncitizen calling for less moderation, after all, has no comparable reason for concern under the policy.

Such action lies at the core of viewpoint discrimination. “At its most basic, the test for viewpoint discrimination is whether — within the relevant subject category — the government has singled out a subset of messages for disfavor based on the views expressed.” Matal v. Tam, 582 U.S. 218, 248 (2017) (Kennedy, J., concurring in part and concurring in the judgment). The Government calls the disfavored side “censorship,” but a label cannot vindicate the cause. See Chiles v. Salazar, 146 S. Ct. 1010, 1023–24 (2026); NAACP v. Button, 371 U.S. 415, 429 (1963) (Government “cannot foreclose the exercise of constitutional rights by mere labels”). Much of American political debate consists of disagreement over whether a practice is liberty or regulation, safety or suppression, accountability or censorship. The First Amendment does not

41

permit officials to resolve that dispute by attaching legal burdens to the side they condemn. See, e.g., Matal, 582 U.S. at 243 (plurality opinion) (“[g]iving offense is a viewpoint” and cannot serve as neutral basis for disfavoring speech); Iancu v. Brunetti, 588 U.S. 388, 393–94 (2019). Nor does it allow the Government to give one side of a disputed public question an official advantage. As the Circuit has put it, “It is antithetical to a free society for the government to give one side of a debatable public question an advantage in expressing its views to the people.” Frederick Douglass Found., Inc. v. District of Columbia, 82 F.4th 1122, 1141 (D.C. Cir. 2023) (quotation marks omitted).

The Government’s conduct-not-speech argument does little to solve the problem. See Opp. at 26. To be sure, the Government may sometimes regulate conduct even when that conduct is carried out through speech, and, as explained below, see infra Section III.A.4.d, it has a legitimate interest in responding when foreign officials use sovereign power to suppress protected expression in the United States. But Defendants cannot recast the challenged applications as conduct merely by calling the underlying work “censorship.” Chiles, 146 S. Ct. at 1023–24; Button, 371 U.S. at 429. As just described, the policy’s demonstrated operation treats reports, petitions, research, and advocacy favoring content moderation as the basis for immigration consequences. That is not incidental expression attending a separately regulable act; the expression is the act, and the policy encompasses it for what it conveys. Cf. Holder v. Humanitarian L. Project, 561 U.S. 1, 27–28 (2010) (rejecting contention that “the only thing actually at issue here is conduct” — and not speech — where “the conduct triggering coverage under the statute consists of communicating a message”).

The ordinary consequence of that conclusion is straightforward. Viewpoint discrimination is “an egregious form of content discrimination.” Rosenberger, 515 U.S. at 829.

42

It is presumptively unconstitutional, “and governments in this country must nearly always abstain from it.” Chiles, 146 S. Ct. at 1021 (quotation marks omitted). Defendants answer that this ordinary rule gives way to immigration deference. In their view, because the policy is implemented through visa restrictions and deportability determinations, the Court may ask only whether the Secretary has supplied a facially legitimate and bona fide reason or at most whether the policy is rationally related to a foreign-policy objective. See Opp. at 22–24; Dep’t of State v. Muñoz, 602 U.S. 899, 908 (2024); Kleindienst v. Mandel, 408 U.S. 753, 769–70 (1972); Trump v. Hawaii, 585 U.S. 667, 704–05 (2018).

Whatever deference applies to exclusion decisions and broad entry restrictions, that line of authority does not resolve a unified policy that burdens a domestic organization’s own expressive conduct and association and also reaches noncitizens already living in the United States, including lawful permanent residents subject to deportability determinations. See Zadvydas v. Davis, 533 U.S. 678, 693 (2001) (“The distinction between an alien who has effected an entry into the United States and one who has never entered runs throughout immigration law.”). Nor does resolving CITR’s challenge require the Court to review any individual visa denial, order any person admitted, or set aside any particular removal order.

In any event, the policy is unlikely to survive even Defendants’ formulation. The Government contends that it must offer a “facially legitimate and bona fide” reason in support of its policy. Muñoz, 602 U.S. at 908 (emphasis added) (quotation marks omitted). Protecting Americans from foreign officials who use sovereign power to suppress protected expression in the United States is in the Government’s interest. But the record does not show that the policy serves only that end. It instead brands a range of private expressive and platform-governance activity as “censorship,” without identifying any foreign-sovereign power that those actors

43

exercised or helped exercise. The Government cannot make protected private expression a facially legitimate and bona fide basis for immigration consequences simply by placing it under the capacious and contested label of “censorship.” Cf. Abourezk v. Reagan, 592 F. Supp. 880, 887 (D.D.C. 1984) (entry may not be denied “solely on account of the content of speech”), vacated and remanded on other grounds by 785 F.2d 1043 (D.C. Cir. 1986).

Nor can Defendants claim the still more forgiving standard of Trump v. Hawaii.

Rational-basis review applied there because the entry restriction was neutral on its face and “expressly premised on legitimate purposes,” resting as it did on a documented, worldwide review of foreign governments’ information-sharing practices. See 585 U.S. at 706–07. The policy here inverts both features. Its operative instruments are not neutral: the Cable’s enumerated categories and the Department’s announced enforcement rationales key immigration consequences to work favoring content moderation. The record, furthermore, contains no comparable predicate review.

The Court therefore need not choose between strict scrutiny and Defendants’ more deferential standard at this stage. Under either approach, CITR is likely to show that the policy’s demonstrated application to private pro-moderation research, advocacy, and association — lacking foreign-sovereign coercive authority — is viewpoint based and unconstitutional. The remaining question is whether those unconstitutional applications are substantial when measured against the policy’s plainly legitimate sweep.

d. The Legitimate Sweep The above conclusion does not mean that the policy has no legitimate sweep. It does, but not as broad as Defendants suggest. Whatever legitimacy the policy carries at its foreign- sovereign core thins as it reaches the many applications that have nothing to do with such power.

44

To begin, the Government has a legitimate interest in protecting Americans’ speech from foreign officials who use sovereign power to suppress it. Cf. TikTok Inc. v. Garland, 604 U.S. 56, 75 (2025) (affording “substantial respect” to Government’s informed judgment about foreign-adversary risks to major platform). The May Memo’s examples largely describe that kind of foreign-sovereign coercion: threats of arrest for activity on American platforms, fines or detention for failing to meet censorship demands, and demands for private user data to police compliance with content-moderation rules. See May Mem. at ECF p. 2. Applied to a foreign official who uses public power to force an American platform to suppress protected expression here, the policy would rest on the serious interest the Government invokes. The July 2025 Brazilian visa revocations, whatever their ultimate merits, at least fit that description: they were aimed at officials said to wield judicial authority against an American platform. See Compl., ¶ 41.

The immigration statutes confirm that protected expression may bear on that kind of foreign-policy judgment. But they also show that Congress subjected such judgments to a speech-protective constraint. A noncitizen generally “shall not be excludable or subject to restrictions or conditions on entry . . . because of the alien’s past, current, or expected beliefs, statements, or associations” that “would be lawful within the United States.” 8 U.S.C. § 1182(a)(3)(C)(iii); see id., § 1227(a)(4)(C)(ii) (incorporating that limitation for removal). Congress allowed the Secretary to act despite that rule only by personally determining that the particular noncitizen’s admission or presence would compromise a compelling United States foreign-policy interest and by notifying designated committees of Congress. See id., §§ 1182(a)(3)(C)(iii)–(iv), 1227(a)(4)(C)(ii). CITR does not deny that authority. Its point is that individualized determinations may apply a valid foreign-policy ground; they cannot transform a

45

policy-level premise that treats one side of a domestic speech debate, without an articulated foreign-sovereign nexus, into part of the policy’s legitimate sweep.

With the policy’s valid core thus identified, the two sides of Moody’s comparison take shape. On one side of the ledger lies action against foreign officials, and perhaps private actors working with them, who use foreign governmental power to suppress protected expression in the United States. The entry component of the policy is part of the field the Court must survey, and applications to foreign officials abroad who coerce American platforms belong to that core. The Court does not minimize that interest or suggest that the policy is invalid in all its applications. But the column holds only what the record connects to that interest. The legitimate sweep must be drawn “from the text of [the policy] and from actual fact.” Hicks, 539 U.S. at 122 (quotation marks omitted). Beyond the May Memo’s illustrative examples and an enforcement action against a single Brazilian justice, little else shows the policy trained on foreign-sovereign censorship. The action against Thierry Breton is at most a closer case: the Department cited a 2024 letter he sent to X while serving as a European Commissioner concerning that platform’s obligations under European law, though it acted against him only after he had left office. See Compl., ¶ 56. The other demonstrated applications, in contrast, reach well beyond the boundary Defendants describe. See supra Section III.A.4.b. As recounted, the Government could not connect those enforcement examples to that professed boundary when asked at oral argument. See Hr’g Tr. at 30:22–31:16. The Court may not supply the connection on Defendants’ behalf: the same rule that forbids speculating a policy into invalidity forbids speculating one into validity. See Wash. State Grange, 552 U.S. at 449–50.

That the column runs thin is confirmed by, though it does not depend on, State’s own assessment of the asserted foreign-sovereign threat. The Department reportedly examined

46

whether European regulators were using the Digital Services Act to censor American speech and found “no evidence that Member States of the European Union are overreaching the [Digital Services Act] to censor and criminalize online content.” ECF No. 11-17 (Wash. Post Article) at ECF p. 1 (quotation marks omitted); see also Pl. Reply at 13. The Court does not second-guess the Executive’s predictive judgments in the abstract. But neither foreign-affairs deference nor immigration deference requires the Court to accept the Government’s bare characterization of protected private expression as foreign censorship.

On the other side of the ledger lies the bulk of what the policy has been shown to do in this record. The December Cable is not confined to foreign officials, sovereign compulsion, or cooperation with a foreign-censorship regime. It directs scrutiny toward work in the platform- governance field — including efforts to combat false narratives, fact checking, moderation, compliance, and trust-and-safety functions. See NPR Article at ECF p. 3. Nor is the Cable a solitary application. It is a standing, programmatic directive that applies the policy’s criteria to every covered visa applicant whose work history touches the enumerated fields. The December enforcement actions touching CITR’s community likewise rested on reports, ratings, advocacy, petitions, interviews, and nonprofit leadership, with no demonstration of foreign-sovereign coercive authority. See Compl., ¶¶ 56–60. And, as explained above, the Department warned that the policy could reach still more, leaving ordinary platform-governance work within its potential grasp. See supra Section III.A.4.b.

Each of those applications does more than exclude or remove a single worker. The work the policy targets is collaborative and institutional by nature: it is carried out through research coalitions like CITR, universities, and nonprofit organizations that study the platforms, publish what they find, and advocate for reform. A policy that targets such work on the basis of

47

viewpoint thus impairs those institutions’ own protected ability to conduct and publish research, to staff and convene their projects, and to associate with the colleagues on whom the work depends. CITR’s record makes the impairment concrete — contributors lost to its reporting, participation drained from its convenings, association with the Coalition made costly — but the burden falls on any institution in the field whose researchers, employees, or members the policy covers.

Measured against one another, the policy’s legitimate applications ultimately do not carry the day. The mismatch between Defendants’ asserted interest and the policy’s demonstrated operation is stark. See Ams. for Prosperity Found. v. Bonta, 594 U.S. 595, 612 (2021) (faulting “dramatic mismatch . . . between the interest that the [Government] seeks to promote and the . . . regime”); cf. Pursuing Am.’s Greatness v. FEC, 831 F.3d 500, 510–11 (D.C. Cir. 2016) (explaining that speech restrictions must rest on more than “anecdote and supposition”). The defect identified above is not a feature of any one application; it is the policy’s selection criteria itself, and it travels wherever the policy does — into visa screening, exclusion, and removal alike. The policy’s legitimate applications, by contrast, remain episodic and largely undemonstrated. Whatever arithmetic might refine the comparison, the overbreadth inquiry asks whether a measure “prohibits a substantial amount of protected speech relative to its plainly legitimate sweep,” Hansen, 599 U.S. at 770, and a policy that selects its targets by an unconstitutional criteria, while its lawful uses remain occasional and largely unproven, answers that question.

On balance, then, the unconstitutional applications are likely substantial enough to carry CITR’s burden. The Court concludes no more than that. It does not hold that Defendants are powerless when a foreign official wields sovereign authority to suppress protected expression in

48

the United States, and it does not pass on the lawfulness of any individual visa denial or removal determination. It holds only that CITR is likely to show that the policy, as expanded in December and applied since, sweeps into the category of “foreign censorship” a substantial measure of the research, reporting, advocacy, and association through which CITR and institutions like it carry out their protected work, and it does so on the basis of viewpoint, in violation of the First Amendment. CITR is therefore likely to succeed on that claim.

5. Administrative Procedure Act The merits of Plaintiff’s APA count require little additional work. The statute directs courts to “hold unlawful and set aside agency action” that is “contrary to constitutional right.” 5 U.S.C. § 706(2)(B). As explained above, that provision encapsulates CITR’s First Amendment claim. See supra Section III.A.4. Having concluded that the Coalition is likely to succeed on that front, the Court accordingly finds that Plaintiff is likely to succeed under § 706(2)(B) as well and need not sort through the remaining APA theories.

That, however, does not end the matter. Section 705 authorizes interim relief only when a court is engaged in judicial review of agency action, and § 704 makes review available only for “final agency action for which there is no other adequate remedy in a court.” 5 U.S.C. §§ 704– 705. The Government contests both finality and adequate remedy, see Opp. at 29–31, so the Court takes each in turn.

a. Final Agency Action Finality turns on two conditions, both of which must be satisfied. “First, the action must mark the consummation of the agency’s decisionmaking process”; and “second, the action must be one by which rights or obligations have been determined, or from which legal consequences will flow.” Bennett v. Spear, 520 U.S. 154, 177–78 (1997) (quotation marks omitted); accord

49

U.S. Army Corps of Eng’rs v. Hawkes Co., 578 U.S. 590, 597 (2016). The inquiry is “pragmatic,” Hawkes, 578 U.S. at 599 (quotation marks omitted), and a central consideration is “the actual legal effect (or lack thereof) of the agency action.” Nat’l Mining Ass’n v. McCarthy, 758 F.3d 243, 252 (D.C. Cir. 2014).

The Government contests both conditions, and its two arguments share a premise: that the policy is a mere enforcement framework, the non-final precursor to the individualized visa and removal determinations in which any rights are actually fixed. See Opp. at 29–30. On consummation, the Government says that the policy announces only a general objective and leaves every operative choice for later; on legal consequences, it maintains that whatever befalls any noncitizen flows from those downstream determinations, not from the policy itself. Neither convinces.

Take consummation first. The policy was not “merely tentative or interlocutory.”

Bennett, 520 U.S. at 178. That conclusion follows from the Court’s earlier finding that the December actions did not stand apart from the policy but carried it forward. See supra Section III.A.1. In May, the Secretary “announc[ed] a new visa restriction policy” that applied to foreign nationals deemed “complicit in censoring Americans.” ECF No. 11-18 (Rubio X Post of May 28, 2025). In December, the Department issued a cable invoking that policy and directing consular officers to scrutinize H-1B applicants for content-moderation work. See NPR Article at ECF pp. 2–3. The Department then grounded its December enforcement actions in the same policy, explaining that Ahmed’s deportability determination was “[c]onsistent with those foreign policy objectives articulated in the May 2025 3C policy.” Dec. Action Mem. at 5. That sequence reflects a consummated agency position: the Department had decided that work in content moderation and related fields could count as complicity in censorship and trigger immigration

50

consequences. Cf. Ciba-Geigy Corp. v. EPA, 801 F.2d 430, 436–37 (D.C. Cir. 1986) (finding finality where agency “publicly articulate[d] an unequivocal position . . . and expects regulated entities to alter their primary conduct to conform to that position”).

The second condition is satisfied because the policy has already produced definite legal consequences: visas revoked and a noncitizen rendered deportable, each a “direct and appreciable legal consequence[].” Cal. Cmtys. Against Toxics v. EPA, 934 F.3d 627, 637 (D.C. Cir. 2019). That is so even though the policy operates through individualized determinations. This Court’s decision in R.I.L-R v. Johnson, 80 F. Supp. 3d 164 (D.D.C. 2015), is the closest analogue. There, the plaintiffs challenged ICE’s consideration of deterrence of mass migration as an allegedly impermissible factor in custody determinations; the Government responded, as it does here, that what plaintiffs called a “policy” was instead only a “generalized agency decision- making process” left to case-by-case judgment. Id. at 184 (quotation marks omitted). This Court disagreed, holding that the plaintiffs had attacked “particularized agency action” — the consideration of that factor — and that the policy had “profound and immediate consequences” for those detained under it. Id. (emphasis omitted); see also Aracely R. v. Nielsen, 319 F. Supp. 3d 110, 138–39 (D.D.C. 2018) (finding deterrence policy reviewable final agency action notwithstanding individualized parole decisions). CITR likewise challenges the policy-level criteria the Department has directed its officers to apply, not the outcome of any one decision.

In any event, the Government’s characterization is difficult to reconcile with its own words. Beyond the documents’ own references to “the May 2025 3C policy,” Dec. Action Mem. at 5, the Department’s Under Secretary told a congressional committee that Ahmed was “one of the targets of our recent 3C sanctions policy.” DeCell Suppl. Decl., ¶ 6. Those statements are not conclusive, but they are probative of whether the Department understood itself to be applying

51

a cohesive policy. Having “repeatedly stated that there is a policy,” the Department “cannot, now, have [its] cake and eat it too.” Amadei v. Nielsen, 348 F. Supp. 3d 145, 165 (E.D.N.Y. 2018). It cannot publicly describe the December actions as enforcement of a 3C sanctions policy and then disclaim that policy in litigation.

The policy thus marks the consummation of the Department’s decisionmaking and carries direct legal consequences.

b. No Other Adequate Remedy The Government next contends that APA review is unavailable because U.S.-based noncitizens may raise their objections in removal proceedings and then in a petition for review. See Opp. at 30. That position largely repackages the channeling arguments already addressed. The INA provisions that Defendants invoke provide review of individual immigration decisions; they do not provide CITR a forum for its own challenge to the policy. See supra Section III.A.3. For the same reason, they do not furnish an adequate alternative remedy under § 704.

Section 704 forecloses APA review only when Congress has supplied another remedy in court that is adequate for the claim and injury at issue. See Bowen v. Massachusetts, 487 U.S. 879, 903 (1988); El Rio Santa Cruz Neighborhood Health Ctr., Inc. v. HHS, 396 F.3d 1265, 1270–75 (D.C. Cir. 2005). The alternative need not be identical, but it must be of the “same genre.” Garcia v. Vilsack, 563 F.3d 519, 522 (D.C. Cir. 2009) (quotation marks omitted). The petition-for-review process is not that. CITR holds no visa, is not subject to removal, and cannot obtain review from an immigration judge or court of appeals by petitioning from someone else’s order of removal. Nor would such proceedings address the injuries on which CITR’s standing rests: chilled participation in its reports and convenings, loss of public association under its banner, disruption of its research and advocacy work, and diversion of organizational resources.

52

That mismatch is the same one courts recognize when applying the INA’s channeling provisions. Section 1252(b)(9) does not funnel every challenge touching removal into a petition for review; courts distinguish review of individual immigration determinations from collateral challenges to the policies and practices used to make them. See McNary v. Haitian Refugee Ctr., Inc., 498 U.S. 479, 492–97 (1991); Gen. Elec. Co. v. Jackson, 610 F.3d 110, 125–27 (D.C. Cir. 2010) (explaining that McNary preserves “general collateral challenges to unconstitutional practices and policies”); Jafarzadeh v. Duke, 270 F. Supp. 3d 296, 306–10 (D.D.C. 2017) (holding that § 1252(b)(9) did not bar challenge to screening process because such claims are “wholly collateral to the statute’s review provisions”); O.A., 404 F. Supp. 3d at 131–33 (declining to treat facial challenge to generally applicable asylum rule as challenge to individual removal order). The same reasoning applies here. CITR is not using the APA to bypass review of any particular removal order; it challenges the policy itself and seeks relief for injuries that the INA review scheme would not redress.

That is particularly true given the remedial posture. The Government itself contends that § 1252(f)(1) bars injunctive relief on CITR’s freestanding First Amendment claim. See Opp. at 38–39. If that argument has force, the § 705 stay CITR seeks is not duplicative of another adequate remedy; it is the means Congress supplied for preserving the status quo while a court reviews final agency action. See Am. Ass’n of Univ. Professors, 780 F. Supp. 3d at 386 (observing that if plaintiffs cannot obtain First Amendment injunction, APA stay may be their only effective remedy).

B. Irreparable Harm The Court at last leaves likelihood of success behind, though the question of irreparable harm does not require it to travel far. It has long been recognized that “[t]he loss of First

53

Amendment freedoms, for even minimal periods of time, unquestionably constitutes irreparable injury.” Elrod v. Burns, 427 U.S. 347, 373 (1976) (plurality opinion); see also Pursuing Am.’s Greatness, 831 F.3d at 511. To be sure, that principle does not relieve CITR of showing that its protected interests are “threatened or in fact being impaired.” Chaplaincy of Full Gospel Churches, 454 F.3d at 301 (quotation marks omitted). Yet the Coalition has done so. As canvassed at length above, CITR has shown that the Government’s policy has chilled the speech and association on which the organization’s research, publication, and convening work depend. Such “self-censor[ship]” in research and publication decisions is the very sort of First Amendment injury that satisfies irreparable harm. Media Matters for Am. v. Paxton, 732 F. Supp. 3d 1, 29 (D.D.C. 2024), aff’d, 138 F.4th 563 (D.C. Cir. 2025).

Defendants do not seriously contest that proposition. They instead look to the calendar, insisting that CITR waited too long because the Secretary first announced the policy in May 2025. See Opp. at 40. Timing can, of course, matter: an “unexcused delay in seeking extraordinary injunctive relief” may “impl[y] a lack of urgency and irreparable harm.” Fed. Educ. Ass’n v. Trump, 795 F. Supp. 3d 74, 100 (D.D.C. 2025) (quotation marks omitted). That principle, however, has limits. Delay, standing alone, cannot defeat preliminary relief. Gordon v. Holder, 632 F.3d 722, 724–25 (D.C. Cir. 2011).

Defendants also wind their clock too early. CITR challenges a policy whose force became materially clearer in December, when State issued further guidance and then enforced the policy against named individuals, including leaders of organizations in CITR’s orbit. That is when the threat ceased to look merely announced and began to look operational. The resulting chill is no stale grievance dredged up for this litigation; it deepens each day the policy remains in force. CITR’s timing therefore does not undercut its showing of irreparable harm.

54

C. Balance of Equities and Public Interest The remaining factors point in the same direction, though not overwhelmingly. The balance of equities and the public interest “merge when the Government is the opposing party.” Nken, 556 U.S. at 435. Defendants invoke substantial interests: enforcing immigration laws, conducting foreign affairs, and avoiding undue judicial interference with the Executive in those domains. The Court is mindful of those concerns, particularly where they implicate the “sensitive and weighty interests of national security and foreign affairs.” TikTok Inc. v. Trump, 507 F. Supp. 3d 92, 114 (D.D.C. 2020) (quoting Humanitarian L. Project, 561 U.S. at 33–34).

But those interests do not answer the relevant question at Defendants’ chosen level of generality. The question is not whether the Government may enforce the immigration laws or make foreign-policy judgments. It plainly may. The question is whether the public interest favors allowing Defendants, while this case proceeds, to enforce a policy likely to violate the First Amendment. It does not. “[T]here is always a strong public interest in the exercise of free speech rights otherwise abridged by an unconstitutional regulation.” Pursuing Am.’s Greatness, 831 F.3d at 511. And “[t]he Constitution does not permit [the Government] to prioritize any policy goal over constitutional rights.” Turner v. U.S. Agency for Glob. Media, 502 F. Supp. 3d 333, 386 (D.D.C. 2020) (quotation marks omitted).

Nor would the relief CITR seeks disable Defendants from pursuing their legitimate interests. A stay would not bar the Government from enforcing the immigration laws, making foreign-policy determinations, or acting on grounds independent of protected expression. It would prevent only the application of the challenged policy in a manner likely to violate the First Amendment. The Government “cannot suffer harm from an injunction that merely ends an unlawful practice,” R.I.L-R, 80 F. Supp. 3d at 191 (quotation marks omitted), and “[t]here is

55

generally no public interest in the perpetuation of unlawful agency action.” League of Women Voters of U.S. v. Newby, 838 F.3d 1, 12 (D.C. Cir. 2016). On the contrary, there is “a substantial public interest ‘in having governmental agencies abide by the federal laws that govern their existence and operations.’” Id. (quoting Washington v. Reno, 35 F.3d 1093, 1103 (6th Cir. 1994)). The equities and public interest therefore favor preliminary relief.

D. Final Issues Although all four factors thus favor preliminary relief, the Court’s work is not yet done.

The Government presses two objections to the form that relief should take — one to its scope, the other to the security that should accompany it. See Opp. at 41–42. Plaintiff, for its part, seeks a protective order restricting Defendants’ use of litigation information and barring retaliation. See ECF No. 10 (Protective Order Mot.). The Court begins with the Government’s objections before turning to CITR’s request.

As to the scope of relief, Defendants invoke Trump v. CASA, Inc., 606 U.S. 831 (2025), to argue that any relief must run no further than CITR’s members. See Opp. at 41. They acknowledge that this Circuit has said otherwise as to § 705 stays but ask the Court to treat that statement as nonbinding and mistaken. Id. (citing Make the Rd. N.Y. v. Noem, 2025 WL 3563313, at *34 (D.C. Cir. Nov. 22, 2025) (statement of Millett and Childs, JJ.) (CASA “is not a case about the scope of relief for agency review authorized by the APA.”) (emphasis omitted)). The argument has since lost its purchase, as demonstrated by the more recent — and binding — decision in Make the Road.

CASA addressed whether the Judiciary Act of 1789 gave lower federal courts equitable authority to issue universal injunctions — orders that prohibit the Government from enforcing a law or policy beyond the parties before the court — and held that it did not. See 606 U.S. at

56

837–38. Even so, the Court expressly declined to address the scope of relief the APA affords. Id. at 847 n.10; see also id. at 869 (Kavanaugh, J., concurring).

Make the Road closed the gap. As the Circuit explained, a § 705 stay is not the preliminary form of an injunction but of vacatur, much as a preliminary injunction is the preliminary form of a permanent one. See 2026 WL 1792978, at *12. That lineage marks the difference that matters. An injunction “directs the conduct of a party” with “the backing of [the court’s] full coercive powers.” Id. at *11 (quoting Nken, 556 U.S. at 428). Vacatur and its preliminary form, a stay, do nothing of the kind: they “operate[] on the legal status of the challenged agency action,” suspending its force rather than ordering any party to act. Id. Because CASA limited the equitable authority to enter injunctions running beyond the parties, its reach does not extend to APA relief aimed at agency action.

The second objection fares no better. The Government asks that CITR post security under Federal Rule of Civil Procedure 65(c). See Opp. at 42. That Rule, however, conditions only a “preliminary injunction or a temporary restraining order” on the giving of security. See Fed. R. Civ. P. 65(c). As discussed above, a § 705 stay is not an injunction, and the APA’s stay provision carries no security requirement of its own. Courts have therefore declined to demand a bond as the price of a § 705 stay. See Cabrera v. U.S. Dep’t of Lab., 792 F. Supp. 3d 91, 107 n.3 (D.D.C. 2025); Neguse v. U.S. Immigr. & Customs Enf’t, 813 F. Supp. 3d 45, 100 (D.D.C. 2025); Am. Ass’n of Nurse Pracs. v. McMahon, --- F. Supp. 3d ---, 2026 WL 1826176, at *23 (D.D.C. 2026) (collecting cases). The Government, for its part, points to no decision making a bond a condition of such a stay. The Court will require none here.

That leaves Plaintiff’s Motion for a Protective Order. CITR primarily seeks two forms of protection: a bar against the Government’s using any information “filed or disclosed” in this case

57

for immigration enforcement or investigations, and a broader prohibition on retaliation for participating in this suit, backed by advance notice and judicial intervention before specific immigration or investigatory actions may proceed. See ECF No. 10-2 (Proposed Order), ¶¶ 3–5.

The Coalition’s concern is understandable: members fear both enforcement under the challenged policy and other retaliatory immigration, investigative, or regulatory action. See Protective Order Mot., ¶¶ 2–4. Yet such expansive relief is not warranted on the present record. CITR’s proposed use restriction would bar a host of governmental actions based on information of all kinds elicited in this case, extending even to persons merely “identified through this lawsuit.” Id., ¶ 8. As the Court today stays the policy and Plaintiff has already submitted the evidence supporting such preliminary relief on the public docket, the immediate risk that litigation disclosures will facilitate the challenged enforcement regime is substantially diminished.

The stay does not eliminate CITR’s separate fear of broader retaliation, but it changes the nature of the risk that remains. With the challenged policy suspended, the residual concern rests on the possibility of future unlawful acts rather than any concrete threat now before the Court. That posture stands in contrast to cases in which protective orders addressed a particularized risk arising from the disclosure or use of defined information. See Standing Rock Sioux Tribe v. U.S. Army Corps of Eng’rs, 249 F. Supp. 3d 516, 520–24 (D.D.C. 2017) (protecting specified portions of record from public disclosure based on particularized security risk); U.S. EEOC v. SOL Mexican Grill LLC, 2019 WL 2896933, at *2–6 (D.D.C. June 11, 2019) (barring discovery into immigration status and employment history where such inquiries were largely irrelevant and likely to chill participation). Doe v. Mayorkas, 2021 WL 9036563 (D.D.C. Feb. 8, 2021), is particularly instructive. Presented with a proposed order that would both limit dissemination of

58

the plaintiff’s identifying information and bar its use for removal, the court issued the first form of protection on the basis of privacy, see Doe v. Mayorkas, No. 20-2521, ECF No. 24 (D.D.C. Mar. 1, 2021), but declined the second, holding that a “hypothetical risk of retaliation” could not sustain such prospective relief. Doe v. Mayorkas, 2021 WL 9036563, at *2–4.

CITR’s proposed order errs on the broader end of the spectrum. It would convert an existing prohibition on retaliation into a contempt-backed command; extend that exposure to any adverse action against an undefined population of members and their families as well as the Coalition’s fiscal sponsor; and make the Court a continuing screener of independently initiated government actions — even after this litigation ends and with enforcement turning on “elusive questions of motivation.” Twelve John Does v. District of Columbia, 117 F.3d 571, 579 (D.C. Cir. 1997); see Proposed Order, ¶¶ 3–5, 8. The Court will therefore deny the Motion for a Protective Order without prejudice. Should a more concrete and immediate threat appear in the future, Plaintiff is free to seek more specific relief from the Court. IV. Conclusion CITR has shown that the preliminary-relief factors favor preserving the status quo while this case proceeds. The Court will therefore grant its Motion to the extent it seeks a stay under 5 U.S.C. § 705 and will stay the challenged policy pending resolution of this action. Plaintiff’s separate Motion for a Protective Order will be denied without prejudice. A contemporaneous Order so stating will issue this day.

/s/ James E. Boasberg

JAMES E. BOASBERG

Chief Judge

Date: July 14, 2026