������ýapp�ٷ�

NOTICE: This opinion is subject to formal revision before publication in the United States Reports. Readers are requested to notify the Reporter of Decisions, Supreme Court of the United States, Washington, D. C. 20543, pio@supremecourt.gov, of any typographical or other formal errors.

SUPREME COURT OF THE UNITED STATES

_________________

No. 23–411

_________________

VIVEK H. MURTHY, SURGEON GENERAL, et al., PETITIONERS v. MISSOURI, et al.

on writ of certiorari to the united states court of appeals for the fifth circuit

[June 26, 2024]

Justice Barrett delivered the opinion of the Court.

During the 2020 election season and the COVID–19 pandemic, social-media platforms frequently removed, demoted, or fact checked posts containing allegedly false or misleading information. At the same time, federal officials, concerned about the spread of “misinformation” on social media, communicated extensively with the platforms about their content-moderation efforts.

The plaintiffs, two States and five social-media users, sued dozens of Executive Branch officials and agencies, alleging that they pressured the platforms to suppress protected speech in violation of the First Amendment. The Fifth Circuit agreed, concluding that the officials’ communications rendered them responsible for the private ������ٴڴǰ�����’ moderation decisions. It then affirmed a sweeping preliminary injunction.

The Fifth Circuit was wrong to do so. To establish standing, the plaintiffs must demonstrate a substantial risk that, in the near future, they will suffer an injury that is traceable to a Government defendant and redressable by the injunction they seek. Because no plaintiff has carried that burden, none has standing to seek a preliminary injunction.

I

A

With their billions of active users, the world’s major social-media companies host a “staggering” amount of content on their platforms. Twitter, Inc. v. Taamneh, 598 U.S. 471, 480 (2023). Yet for many of these companies, including Facebook, Twitter, and YouTube, not everything goes.[1] Under their longstanding content-moderation policies, the platforms have taken a range of actions to suppress certain categories of speech. They place warning labels on some posts, while deleting others. They also “demote” content so that it is less visible to other users. And they may suspend or ban users who frequently post content that violates platform policies.

For years, the platforms have targeted speech they judge to be false or misleading. For instance, in 2016, Facebook began fact checking and demoting posts containing misleading claims about elections. Since 2018, Facebook has removed health-related misinformation, including false claims about a measles outbreak in Samoa and the polio vaccine in Pakistan. Likewise, in 2019, YouTube announced that it would “demonetize” channels that promote anti-vaccine messages.

In 2020, with the outbreak of COVID–19, the platforms announced that they would enforce their policies against users who post false or misleading content about the pandemic. As early as January 2020, Facebook deleted posts it deemed false regarding “cures,” “treatments,” and the effect of “physical distancing.” 60 Record on Appeal 19,035 (Record). And it demoted posts containing what it described as “conspiracy theories about the origin of the virus.” Id., at 19,036. Twitter and YouTube began applying their policies in March and May 2020, respectively. Throughout the pandemic, the platforms removed or reduced posts questioning the efficacy and safety of mask wearing and the COVID–19 vaccine, along with posts on related topics.

The platforms also applied their misinformation policies during the 2020 Presidential election season. Facebook, in late 2019, unveiled measures to counter foreign interference campaigns and voter suppression efforts. One month before the election, multiple platforms suppressed a report about Hunter Biden’s laptop, believing that the story originated from a Russian hack-and-leak operation. After the election, the platforms took action against users or posts that questioned the integrity of the election results.

Over the past few years, various federal officials regularly spoke with the platforms about COVID–19 and election-related misinformation. Officials at the White House, the Office of the Surgeon General, and the Centers for Disease Control and Prevention (CDC) focused on COVID–19 content, while the Federal Bureau of Investigation (FBI) and the Cybersecurity and Infrastructure Security Agency (CISA) concentrated on elections.

White House. In early 2021, and continuing primarily through that year, the Director of Digital Strategy and members of the COVID–19 response team interacted with the platforms about their efforts to suppress vaccine misinformation. They expressed concern that Facebook in particular was “one of the top drivers of vaccine hesitancy,” due to the spread of allegedly false or misleading claims on the platform. App. 659–660. Thus, the officials peppered Facebook (and to a lesser extent, Twitter and YouTube) with detailed questions about their policies, pushed them to suppress certain content, and sometimes recommended policy changes. Some of these communications were more aggressive than others. For example, the director of Digital Strategy, frustrated that Facebook had not removed a particular post, complained: “[L]ast time we did this dance, it ended in an insurrection.” Id., at 698. Another official, unhappy with Facebook’s supposed lack of transparency about its vaccine misinformation problems, wrote: “Internally we have been considering our options on what to do about it.” Id., at 657. Publicly, White House communications officials called on the platforms to do more to address COVID–19 misinformation — and, perhaps as motivation, raised the possibility of reforms aimed at the platforms, including changes to the antitrust laws and 47 U. S. C. §230.

Surgeon General. In July 2021, Surgeon General Vivek Murthy issued a health advisory on misinformation. The advisory encouraged platforms to “[r]edesign recommendation algorithms to avoid amplifying misinformation,” “[i]mpose clear consequences for accounts that repeatedly violate platform policies,” and “[p]rovide information from trusted and credible sources to prevent misconceptions from taking hold.” 3 Record 662. At a press conference to announce the advisory, Surgeon General Murthy argued that the platforms should “operate with greater transparency and accountability.” 2 id., at 626. The following year, the Surgeon General issued a “Request for Information,” seeking, among other things, reports on each ������ٴڴǰ���’s “COVID–19 misinformation policies.” Impact of Health Misinformation in the Digital Information Environment in the United States Throughout the COVID–19 Pandemic Request for Information (RFI), 87 Fed. Reg. 12714 (Mar. 7, 2022).

CDC. Like the White House, the CDC frequently communicated with the platforms about COVID–19 misinformation. In early 2020, Facebook reached out to the agency, seeking authoritative information about the virus that it could post on the platform. The following year, the CDC’s communications expanded to other platforms, including Twitter and YouTube. The CDC hosted meetings and sent reports to the platforms, alerting them to misinformation trends and flagging example posts. The platforms often asked the agency for fact checks on specific claims.

FBI and CISA. These agencies communicated with the platforms about election-related misinformation. They hosted meetings with several platforms in advance of the 2020 Presidential election and the 2022 midterms. The FBI alerted the platforms to posts containing false information about voting, as well as pernicious foreign influence campaigns that might spread on their sites. Shortly before the 2020 election, the FBI warned the platforms about the potential for a Russian hack-and-leak operation. Some companies then updated their moderation policies to prohibit users from posting hacked materials. Until mid-2022, CISA, through its “switchboarding” operations, forwarded third-party reports of election-related misinformation to the platforms. These communications typically stated that the agency “w[ould] not take any action, favorable or unfavorable, toward social media companies based on decisions about how or whether to use this information.” 72 Record 23,223.

B

Respondents are two States and five individual social-media users. They were the plaintiffs below, and for the sake of narrative clarity, we will refer to them as “plaintiffs” in this opinion. (Likewise, we will refer to the Government individuals and agencies as “defendants” rather than petitioners.) The individual plaintiffs — three doctors, the owner of a news website, and a healthcare activist — allege that various platforms removed or demoted their COVID–19 or election-related content between 2020 and 2023. The States, Missouri and Louisiana, claim that the platforms have suppressed the speech of state entities and officials, as well as their citizens’ speech.

Though the platforms restricted the plaintiffs’ content, the plaintiffs maintain that the Federal Government was behind it. Acting on that belief, the plaintiffs sued dozens of Executive Branch officials and agencies, alleging that they pressured the platforms to censor the plaintiffs’ speech in violation of the First Amendment. The States filed their complaint on May 5, 2022. The next month, they moved for a preliminary injunction, seeking to stop the defendants from “taking any steps to demand, urge, encourage, pressure, or otherwise induce” any platform “to censor, suppress, remove, de-platform, suspend, shadow-ban, de-boost, restrict access to content, or take any other adverse action against any speaker, content, or viewpoint expressed on social media.” 1 id., at 253. The individual plaintiffs joined the suit on August 2, 2022.

After granting extensive discovery, the District Court issued a preliminary injunction. Missouri v. Biden, 680 F. Supp. 3d 630, 729 (WD La. 2023). The court held that officials at the White House, the Surgeon General’s Office, the CDC, the FBI, and CISA likely “coerced” or “significantly encouraged” the platforms “to such extent that the[ir content-moderation] decision[s] should be deemed to be the decisions of the Government.” Id., at 694 (internal quotation marks omitted). It enjoined those agencies, along with scores of named and unnamed officials and employees, from taking actions “for the purpose of urging, encouraging, pressuring, or inducing in any manner the removal, deletion, suppression, or reduction of content containing protected free speech posted on social-medial platforms.” Missouri v. Biden, 2023 WL 5841935, *1–*2 (WD La., July 4, 2023).[2]

Following a grant of panel rehearing, the Fifth Circuit affirmed in part and reversed in part. Missouri v. Biden, 83 F. 4th 350 (2023). It first held that the individual plaintiffs had Article III standing to seek injunctive relief, reasoning that the social-media companies had suppressed the plaintiffs’ speech in the past and were likely to do so again in the future, id., at 367–369, and that both of these injuries were “traceable to government-coerced enforcement” of the ������ٴڴǰ���’s policies and “redressable by an injunction against the government officials,” id., at 373. The court also concluded that the States had standing, both because the platforms had restricted the posts of individual state officials and because the States have the “right to listen” to their citizens on social media. Id., at 371–372.

On the merits, the Fifth Circuit explained that “a private party’s conduct may be state action if the government coerced or significantly encouraged it.” Id., at 380 (citing Blum v. Yaretsky, 457 U.S. 991, 1004 (1982); emphasis deleted). To identify coercion, it asked whether “the government compelled the [private party’s] decision by . . . intimating that some form of punishment will follow a failure to comply.” 83 F. 4th, at 380. The court explained that the Government significantly encourages a private party’s choice when it exercises “active, meaningful control, whether by entanglement in the party’s decision-making process or direct involvement in carrying out the decision itself.” Id., at 377.[3]

Applying those tests, the Fifth Circuit determined that White House officials, in conjunction with the Surgeon General’s Office, likely both coerced and significantly encouraged the platforms to moderate content. Id., at 388. The court concluded that the same was true for the FBI. Ibid. It held that the CDC and CISA significantly encouraged (but did not coerce) the ������ٴڴǰ�����’ moderation decisions. Id., at 389, 391.

The Fifth Circuit agreed with the District Court that the equities favored the plaintiffs. Id., at 392–394. It then modified the District Court’s injunction to state that the defendants, and their employees and agents, shall not “ ‘coerce or significantly encourage social-media companies to remove, delete, suppress, or reduce, including through altering their algorithms, posted social-media content containing protected free speech.’ ” Id., at 397. The court did not limit the injunction to the platforms that the plaintiffs use or the topics that the plaintiffs wish to discuss, explaining that the harms stemming from the defendants’ conduct “impac[t] every social-media user.” Id., at 398.

The federal agencies and officials applied to this Court for emergency relief. We stayed the injunction, treated the application as a petition for a writ of certiorari, and granted the petition. 601 U. S. ___ (2023).

II

We begin — and end — with standing. At this stage, neither the individual nor the state plaintiffs have established standing to seek an injunction against any defendant. We therefore lack jurisdiction to reach the merits of the dispute.

A

Article III of the Constitution limits the jurisdiction of federal courts to “Cases” and “Controversies.” The “case or controversy” requirement is “ ‘fundamental to the judiciary’s proper role in our system of government.’ ” Raines v. Byrd, 521 U.S. 811, 818 (1997) (quoting Simon v. Eastern Ky. Welfare Rights Organization, 426 U.S. 26, 37 (1976)). Federal courts can only review statutes and executive actions when necessary “to redress or prevent actual or imminently threatened injury to persons caused by . . . official violation of law.” Summers v. Earth Island Institute, 555 U.S. 488, 492 (2009). As this Court has explained, “[i]f a dispute is not a proper case or controversy, the courts have no business deciding it, or expounding the law in the course of doing so.” DaimlerChrysler Corp. v. Cuno, 547 U.S. 332, 341 (2006).

A proper case or controversy exists only when at least one plaintiff “establish[es] that [she] ha[s] standing to sue.” Raines, 521 U. S., at 818; Department of Commerce v. New York, 588 U.S. 752, 766 (2019). She must show that she has suffered, or will suffer, an injury that is “concrete, particularized, and actual or imminent; fairly traceable to the challenged action; and redressable by a favorable ruling.” Clapper v. Amnesty Int’l USA, 568 U.S. 398, 409 (2013) (internal quotation marks omitted). These requirements help ensure that the plaintiff has “such a personal stake in the outcome of the controversy as to warrant [her] invocation of federal-court jurisdiction.” Summers, 555 U. S., at 493 (internal quotation marks omitted).

The plaintiffs claim standing based on the “direct censorship” of their own speech as well as their “right to listen” to others who faced social-media censorship. Brief for Respondents 19, 22. Notably, both theories depend on the ������ٴڴǰ���’s actions — yet the plaintiffs do not seek to enjoin the platforms from restricting any posts or accounts. They seek to enjoin Government agencies and officials from pressuring or encouraging the platforms to suppress protected speech in the future.

The one-step-removed, anticipatory nature of their alleged injuries presents the plaintiffs with two particular challenges. First, it is a bedrock principle that a federal court cannot redress “injury that results from the independent action of some third party not before the court.” Simon, 426 U. S., at 41–42. In keeping with this principle, we have “been reluctant to endorse standing theories that require guesswork as to how independent decisionmakers will exercise their judgment.” Clapper, 568 U. S., at 413. Rather than guesswork, the plaintiffs must show that the third-party platforms “will likely react in predictable ways” to the defendants’ conduct. Department of Commerce, 588 U. S., at 768. Second, because the plaintiffs request forward-looking relief, they must face “a real and immediate threat of repeated injury.” ��’S��� v. Littleton, 414 U.S. 488, 496 (1974); see also Susan B. Anthony List v. Driehaus, 573 U.S. 149, 158 (2014) (“An allegation of future injury may suffice if the threatened injury is certainly impending, or there is a substantial risk that the harm will occur” (internal quotation marks omitted)). Putting these requirements together, the plaintiffs must show a substantial risk that, in the near future, at least one platform will restrict the speech of at least one plaintiff in response to the actions of at least one Government defendant. On this record, that is a tall order.

Before we evaluate the plaintiffs’ different theories, a few preliminaries: The plaintiff “bears the burden of establishing standing as of the time [s]he brought th[e] lawsuit and maintaining it thereafter.” Carney v. Adams, 592 U.S. 53, 59 (2020). She must support each element of standing “with the manner and degree of evidence required at the successive stages of the litigation.” Lujan v. Defenders of Wildlife, 504 U.S. 555, 561 (1992). At the preliminary injunction stage, then, the plaintiff must make a “clear showing” that she is “likely” to establish each element of standing. See Winter v. Natural Resources Defense Council, Inc., 555 U.S. 7, 22 (2008) (emphasis deleted). Where, as here, the parties have taken discovery, the plaintiff cannot rest on “mere allegations,” but must instead point to factual evidence. See Lujan, 504 U. S., at 561 (internal quotation marks omitted).

B

1

The plaintiffs’ primary theory of standing involves their “direct censorship injuries.” They claim that the restrictions they have experienced in the past on various platforms are traceable to the defendants and that the platforms will continue to censor their speech at the behest of the defendants. So we first consider whether the plaintiffs have demonstrated traceability for their past injuries.

Here, a note of caution: If the plaintiffs were seeking compensatory relief, the traceability of their past injuries would be the whole ball game. But because the plaintiffs are seeking only forward-looking relief, the past injuries are relevant only for their predictive value. See ��’S���, 414 U. S., at 495–496 (“Past exposure to illegal conduct” can serve as evidence of threatened future injury but “does not in itself show a present case or controversy regarding injunctive relief ”). If a plaintiff demonstrates that a particular Government defendant was behind her past social-media restriction, it will be easier for her to prove that she faces a continued risk of future restriction that is likely to be traceable to that same defendant. Conversely, if a plaintiff cannot trace her past injury to one of the defendants, it will be much harder for her to make that showing. See Clapper, 568 U. S., at 411. In the latter situation, the plaintiff would essentially have to build her case from scratch, showing why she has some newfound reason to fear that one of the named defendants will coerce her chosen platform to restrict future speech on a topic about which she plans to post — in this case, either COVID–19 or the upcoming election. Keep in mind, therefore, that the past is relevant only insofar as it is a launching pad for a showing of imminent future injury.

The primary weakness in the record of past restrictions is the lack of specific causation findings with respect to any discrete instance of content moderation. The District Court made none. Nor did the Fifth Circuit, which approached standing at a high level of generality. The platforms, it reasoned, “have engaged in censorship of certain viewpoints on key issues,” while “the government has engaged in a years-long pressure campaign” to ensure that the platforms suppress those viewpoints. 83 F. 4th, at 370. The ������ٴڴǰ�����’ “censorship decisions” — including those affecting the plaintiffs — were thus “likely attributable at least in part to the ������ٴڴǰ�����’ reluctance to risk” the consequences of refusing to “adhere to the government’s directives.” Ibid.

We reject this overly broad assertion. As already discussed, the platforms moderated similar content long before any of the Government defendants engaged in the challenged conduct. In fact, the platforms, acting independently, had strengthened their pre-existing content-moderation policies before the Government defendants got involved. For instance, Facebook announced an expansion of its COVID–19 misinformation policies in early February 2021, before White House officials began communicating with the platform. And the platforms continued to exercise their independent judgment even after communications with the defendants began. For example, on several occasions, various platforms explained that White House officials had flagged content that did not violate company policy. Moreover, the platforms did not speak only with the defendants about content moderation; they also regularly consulted with outside experts.

This evidence indicates that the platforms had independent incentives to moderate content and often exercised their own judgment. To be sure, the record reflects that the Government defendants played a role in at least some of the ������ٴڴǰ�����’ moderation choices. But the Fifth Circuit, by attributing every platform decision at least in part to the defendants, glossed over complexities in the evidence.[4]

The Fifth Circuit also erred by treating the defendants, plaintiffs, and platforms each as a unified whole. Our decisions make clear that “standing is not dispensed in gross.” TransUnion LLC v. Ramirez, 594 U.S. 413, 431 (2021). That is, “plaintiffs must demonstrate standing for each claim that they press” against each defendant, “and for each form of relief that they seek.” Ibid. Here, for every defendant, there must be at least one plaintiff with standing to seek an injunction. This requires a certain threshold showing: namely, that a particular defendant pressured a particular platform to censor a particular topic before that platform suppressed a particular plaintiff ’s speech on that topic.

Heeding these conditions is critically important in a sprawling suit like this one. The plaintiffs faced speech restrictions on different platforms, about different topics, at different times. Different groups of defendants communicated with different platforms, about different topics, at different times. And even where the plaintiff, platform, time, content, and defendant line up, the links must be evaluated in light of the ������ٴڴǰ���’s independent incentives to moderate content. As discussed, the platforms began to suppress the plaintiffs’ COVID–19 content before the defendants’ challenged communications started, which complicates the plaintiffs’ effort to demonstrate that each platform acted due to “government-coerced enforcement” of its policies, 83 F. 4th, at 370 (emphasis deleted), rather than in its own judgment as an “ ‘independent acto[r],’ ” Lujan, 504 U. S., at 562. With these factors in mind, we proceed to untangle the mass of the plaintiffs’ injuries and Government communications.

2

The plaintiffs rely on allegations of past Government censorship as evidence that future censorship is likely. But they fail, by and large, to link their past social-media restrictions to the defendants’ communications with the platforms. Thus, the events of the past do little to help any of the plaintiffs establish standing to seek an injunction to prevent future harms.

Louisiana and Missouri. The state plaintiffs devote minimal attention to restriction of their own social-media content, much less to a causal link between any such restriction and the actions of any Government defendant. They refer only to Facebook’s “flagg[ing] . . . and de-boost[ing]” of a Louisiana state representative’s post about children and the COVID–19 vaccine. Brief for Respondents 20; App. 635–636. We need not decide whether an injury to a state representative counts as an injury to the State, because evidence of causation is lacking.[5] The States assert only that in November 2021, Facebook, “as a result of [its] work [with the CDC],” updated its policies “to remove additional false claims about the COVID–19 vaccine for children.” 37 Record 11,457. But they never say when Facebook took action against the official’s post — and a causal link is possible only if the removal occurred after Facebook’s communication with the CDC. There is therefore no evidence to support the States’ allegation that Facebook restricted the state representative pursuant to the CDC-influenced policy.

Jayanta Bhattacharya, Martin Kulldorff, and Aaron Kheriarty. These plaintiffs are doctors who questioned the wisdom of then-prevailing COVID–19 policies, including lockdowns and mask and vaccine mandates. Each faced his first social-media restriction in 2020, before the White House and the CDC entered discussions with the relevant platforms. Plaintiffs highlight restrictions imposed by Twitter and LinkedIn, starting in 2021, on Dr. Kulldorff ’s posts about natural immunity. They also point out that Twitter restricted the visibility of Dr. Kheriarty’s posts about vaccine safety and efficacy, as well as the ethics surrounding vaccine mandates. Attempting to show causation, the plaintiffs emphasize that in January 2022, Facebook reported to White House officials that it had recently demoted one post advocating for natural immunity over vaccine immunity. But neither the timing nor the platforms line up (nor, in Dr. Kheriarty’s case, does the content), so the plaintiffs cannot show that these restrictions were traceable to the White House officials. In fact, there is no record evidence that White House officials ever communicated at all with LinkedIn.

Drs. Bhattacharya and Kulldorff claim that, after disagreeing with the CDC and other federal health officials, they faced a “relentless covert campaign of social-media censorship.” App. 585 (emphasis deleted). They refer to the ������ٴڴǰ�����’ suppression of the Great Barrington Declaration, their coauthored report calling for an end to lockdowns. But their declarations do not suggest that anyone at the CDC was involved; rather, they point to officials at the National Institutes of Health and the NIAID. Those entities are not before us. With nothing else to show, Drs. Bhattacharya, Kulldorff, and Kheriarty have not established a likelihood that their past restrictions are traceable to either the White House officials or the CDC.

Jim Hoft. Both Hoft and his news website, “The Gateway Pundit,” experienced election and COVID–19-related restrictions on various platforms. Hoft tries to demonstrate his standing to sue only the FBI and CISA, which means that only the suppression of his election-related posts is relevant. (As already discussed, the record contains no evidence that either the FBI or CISA engaged with the platforms about the pandemic.) First, Hoft points to the FBI’s role in the ������ٴڴǰ�����’ adoption of hacked-material policies. And he claims that Twitter, in December 2020, censored content about the Hunter Biden laptop story under such a policy. The post was titled: “Where’s Hunter? How is Hunter Celebrating the New Year? New Photos of Hunter Biden Pushing Drugs on Women Emerge.” Hoft’s own declaration reveals that Twitter acted according to its “rules against posting or sharing privately produced/distributed intimate media of someone without their express consent.” Id., at 608. Hoft provides no evidence that Twitter adopted a policy against posting private, intimate content in response to the FBI’s warnings about hack-and-leak operations. Plus, it was Hoft’s brother, Joe Hoft, who posted this tweet; Twitter therefore suspended Joe Hoft’s account. It is unclear why Jim Hoft would have standing to sue for his brother’s injury.

Hoft claims that his content appears on a CISA document tracking posts that various entities had flagged for the platforms as misinformation. The spreadsheet shows that a private entity, the Election Integrity Partnership — not CISA — alerted Twitter to an unidentified article from the Gateway Pundit. And the spreadsheet does not reveal whether Twitter removed or otherwise suppressed that post. This evidence does not support the conclusion that Hoft’s past injuries are likely traceable to the FBI or CISA.

Jill Hines. Of all the plaintiffs, Hines makes the best showing of a connection between her social-media restrictions and communications between the relevant platform (Facebook) and specific defendants (CDC and the White House). That said, most of the lines she draws are tenuous, particularly given her burden of proof at the preliminary injunction stage — recall that she must show that her restrictions are likely traceable to the White House and the CDC.

A healthcare activist, Hines codirects “Health Freedom Louisiana,” a group that advocated against COVID–19 mask and vaccine mandates. In October 2020 — before the start of communications with the White House and the bulk of communications with the CDC — Facebook began to reduce the reach of Hines’ and Health Freedom’s pages. Hines tries to connect Facebook’s subsequent actions against her to both the White House officials and the CDC.

First, Facebook “deplatformed” (i.e., deleted) one of Health Freedom’s groups in July 2021. The last post in the group asked members to contact state legislators about health freedom legislation. Three months earlier, a White House official sent Facebook several “suggestions” that were “circulating around the building and informing thinking,” including that the platform should “end group recommendations for groups with a history of COVID–19 or vaccine misinformation.” 54 Record 16,870–16,871. A week later, Facebook replied that it had “already removed all health groups from our recommendation feature.” App. 716. It is hard to know what to make of this. Facebook reported that it had already acted, which tends to imply that Facebook made its decision independently of the White House. Moreover, Facebook and the White House communicated about removing groups from recommendation features, not deleting them altogether — further weakening the inference that Facebook was implementing White House policy rather than its own.[6]

Next, in April 2023, Facebook gave Hines a warning after she reposted content from Robert F. Kennedy, Jr. Two years earlier, White House officials had pushed Facebook to remove the accounts of the “disinformation dozen,” 12 people (including Kennedy) supposedly responsible for a majority of COVID–19-related misinformation. Hines tries to link the warning she received to this earlier White House pressure. Again, though, the link is weak. There is no evidence that the White House asked Facebook to censor every user who reposts a member of the disinformation dozen, nor did Facebook change its policies to do so. Facebook’s 2023 warning to Hines bears only a tangential relationship to the White House’s 2021 directive to Facebook.

Hines traces her remaining restrictions to the CDC. Beginning in October 2020, Facebook fact checked Hines’ posts about pregnant women taking the COVID–19 vaccine, along with posts including data from the Vaccine Adverse Event Reporting System (VAERS). And in March 2021, the CDC flagged several misinformation trends for Facebook, including claims related to pregnancy and VAERS data. Because Hines does not provide dates for the fact checks, we cannot know whether the CDC could be responsible.

In May 2022, Facebook restricted Hines’ account for posting an article discussing increased rates of myocarditis in teenagers following vaccination. A little over a year earlier, the CDC warned Facebook against claims of “unsubstantiated links to new [vaccine] side effects,” including “ ‘irritab[ility],’ ” “ ‘auto-immune issues, infertility,’ ” and “ ‘neurological damage including lowered IQ.’ ” 54 Record 17,042–17,043 (emphasis deleted). There is no evidence that the CDC ever listed myocarditis as an unsubstantiated side effect — but because it is an alleged side effect, it at least falls under the same umbrella as the CDC’s communication. Health Freedom’s February 2023 violation, by contrast, was for posting that vaccine manufacturers would not compensate those with vaccine-related injuries — a topic that bears little resemblance to the content that the CDC flagged.

In April 2023, Hines received violations for posts about children and the vaccine. In November 2021, Facebook worked with the CDC to update its policies to remove additional false claims including that “ ‘the COVID vaccine is not safe for kids.’ ” 37 id., at 11,457. It is not clear that either of Hines’ posts violated the CDC-influenced policy against false claims related to children and the vaccine. One simply referred to the World Health Organization’s COVID–19 vaccine recommendations for children, and the other discussed the role of children within the “predatory” pharmaceutical industry. App. 789–790. Given the loose match between the policy and the posts, it is hard to call it “likely” that Facebook was enforcing the CDC’s preferences rather than its own.[7]

With one or two potentially viable links, Hines makes the best showing of all the plaintiffs. Still, Facebook was targeting her pages before almost all of its communications with the White House and the CDC, which weakens the inference that her subsequent restrictions are likely traceable to “government-coerced enforcement” of Facebook’s policies, 83 F. 4th, at 370 (emphasis deleted), rather than to Facebook’s independent judgment.[8] Even assuming, however, that Hines has eked out a showing of traceability for her past injuries, the past is relevant only insofar as it predicts the future. And this weak record gives her little momentum going forward.

3

To obtain forward-looking relief, the plaintiffs must establish a substantial risk of future injury that is traceable to the Government defendants and likely to be redressed by an injunction against them. To carry that burden, the plaintiffs must proffer evidence that the defendants’ “allegedly wrongful behavior w[ould] likely occur or continue.” Friends of the Earth, Inc. v. Laidlaw Environmental Services (TOC), Inc., 528 U.S. 167, 190 (2000). At the preliminary injunction stage, the plaintiffs must show that they are likely to succeed in carrying that burden. See Winter, 555 U. S., at 22. But without proof of an ongoing pressure campaign, it is entirely speculative that the ������ٴڴǰ�����’ future moderation decisions will be attributable, even in part, to the defendants.

The plaintiffs treat the defendants as a monolith, claiming broadly that “ ‘the governmen[t]’ ” continues to communicate with the platforms about “ ‘content-moderation issues.’ ” Brief for Respondents 29 (quoting 83 F. 4th, at 369). But we must confirm that each Government defendant continues to engage in the challenged conduct, which is “coercion” and “significant encouragement,” not mere “communication.” Plus, the plaintiffs have only explicitly identified an interest in speaking about COVID–19 or elections — so the defendants’ discussions about content-moderation issues must focus on those topics.

We begin with the plaintiffs who have not pointed to any past restrictions likely traceable to the Government defendants. This failure to establish traceability for past harms — which can serve as evidence of expected future harm — “substantially undermines [the plaintiffs’] standing theory.” Clapper, 568 U. S., at 411. These plaintiffs (i.e., everyone other than Hines) are thus particularly ill suited to the task of establishing their standing to seek forward-looking relief.

Take Hoft, the only plaintiff who has expressed interest in speaking about elections (and thus the only plaintiff with potential standing to sue the FBI and CISA). The FBI’s challenged conduct was ongoing at the time of the complaint, as the agency worked with the platforms during the 2022 midterm election season. Still, Hoft must rely on a “speculative chain of possibilities” to establish a likelihood of future harm traceable to the FBI. Id., at 414. Hoft’s future posts (presumably about the 2024 Presidential election) must contain content that falls within a misinformation trend that the FBI has identified or will identify in the future. The FBI must pressure the platforms to remove content within that category. The platform must then suppress Hoft’s post, and it must do so at least partly in response to the FBI, rather than in keeping with its own content-moderation policy. Hoft cannot satisfy his burden with such conjecture. CISA, meanwhile, stopped switchboarding in mid-2022, and the Government has represented that it will not resume operations for the 2024 election. Especially in light of his poor showing of traceability in the past, Hoft has failed to demonstrate likely future injury at the hands of the FBI or CISA — so the injunction against those entities cannot survive.

The doctors and the state plaintiffs, who focus on COVID–19 content, have a similarly uphill battle vis-à-vis the White House, the Surgeon General’s Office, and the CDC. Hines, with her superior showing on past harm, is in a slightly better position to demonstrate likely future harm at the hands of these defendants. Still, she has not shown enough.

Starting with the White House and Surgeon General’s Office, the vast majority of their public and private engagement with the platforms occurred in 2021, when the pandemic was still in full swing. By August 2022, when Hines joined the case, the officials’ communications about COVID–19 misinformation had slowed to a trickle. Publicly, the White House Press Secretary made two statements in February and April 2022. First, she said that the platforms should continue “call[ing] out misinformation and disinformation.” 3 Record 758. Two months later, she spoke generally about §230 and antitrust reform, but did not mention content moderation or COVID–19 misinformation. In March 2022, the Surgeon General issued a voluntary “Request for Information” from the platforms about their misinformation policies.[9]

Privately, Facebook sent monthly “Covid Insights” reports to officials in the White House and the Surgeon General’s Office, at least until July 2022. These reports contained information about the top 100 vaccine-related posts in the United States, including whether Facebook took action against any of them. In June, Facebook asked if it should continue sending these reports, as it had stopped seeing “problematic vaccine related” content in the top posts. 50 id., at 15,645–15,646. The official replied that, though he would “normally say we are good to discontinue,” the reports would be helpful “as we start to ramp up . . . vaccines” for children under five. Id., at 15,645. The record contains no other evidence of private contact with respect to COVID–19 misinformation.

On this record, it appears that the frequent, intense communications that took place in 2021 had considerably subsided by 2022. (Perhaps unsurprisingly, given the changed state of the pandemic.) It is thus very difficult for Hines to show that she faces future harm that is traceable to officials in the White House and the Surgeon General’s Office. Recall the Fifth Circuit’s reasoning regarding traceability for past harms: In the face of a governmental “pressure campaign,” the “������ٴڴǰ�����’ censorship decisions were likely attributable at least in part to [their] reluctance to risk the adverse legal or regulatory consequences that could result from a refusal to adhere to the government’s directives.” 83 F. 4th, at 370. But in the months leading up to this suit, these officials issued no directives and threatened no consequences. They only asked for information about the most popular vaccine-related posts. Hines does not allege that her content has fallen, or is likely to fall, in that category.

In these circumstances, Hines cannot rely on “the predictable effect of Government action on the decisions of third parties”; rather, she can only “speculat[e] about the decisions of third parties.” Department of Commerce, 588 U. S., at 768. It is “no more than conjecture” to assume that Hines will be subject to White House-induced content moderation. Los Angeles v. Lyons, 461 U.S. 95, 108 (1983). Hines (along with the other plaintiffs) has therefore failed to establish a likelihood of future injury traceable to the White House or the Surgeon General’s Office. Likewise, the risk of future harm traceable to the CDC is minimal. The CDC stopped meeting with the platforms in March 2022. Thereafter, the platforms sporadically asked the CDC to verify or debunk several claims about vaccines. But the agency has not received any such message since the summer of 2022.[10]

The plaintiffs’ counterarguments do not persuade. First, they argue that they suffer “continuing, present adverse effects” from their past restrictions, as they must now self-censor on social media. ��’S���, 414 U. S., at 496. But the plaintiffs “cannot manufacture standing merely by inflicting harm on themselves based on their fears of hypothetical future harm that is not certainly impending.” Clapper, 568 U. S., at 416. And as we explained, the plaintiffs have not shown that they are likely to face a risk of future censorship traceable to the defendants. Indeed, even before the defendants entered the scene, the plaintiffs “had a similar incentive to engage in” self-censorship, given the ������ٴڴǰ�����’ independent content moderation. Id., at 417. So it is “difficult to see how” the plaintiffs’ self-censorship “can be traced to” the defendants. Ibid.

Second, the plaintiffs and the dissent suggest that the platforms continue to suppress their speech according to policies initially adopted under Government pressure. Post, at 21. That may be true. But the plaintiffs have a redressability problem. “To determine whether an injury is redressable,” we “consider the relationship between ‘the judicial relief requested’ and the ‘injury’ suffered.” California v. Texas, 593 U.S. 659, 671 (2021). The plaintiffs assert several injuries — their past social-media restrictions, current self-censorship, and likely social-media restrictions in the future. The requested judicial relief, meanwhile, is an injunction stopping certain Government agencies and employees from coercing or encouraging the platforms to suppress speech. A court could prevent these Government defendants from interfering with the ������ٴڴǰ�����’ independent application of their policies. But without evidence of continued pressure from the defendants, it appears that the platforms remain free to enforce, or not to enforce, those policies — even those tainted by initial governmental coercion. The platforms are “not parties to the suit, and there is no reason they should be obliged to honor an incidental legal determination the suit produced.” Lujan, 504 U. S., at 569 (plurality opinion); see also Haaland v. Brackeen, 599 U.S. 255, 293–294 (2023).

Indeed, the available evidence indicates that the platforms have enforced their policies against COVID–19 misinformation even as the Federal Government has wound down its own pandemic response measures. For instance, Hines reports that Facebook imposed several restrictions on her vaccine-related posts in the spring of 2023. Around the same time, in April 2023, President Biden signed a joint resolution that ended the national COVID–19 emergency. See Pub. L. 118–3, 137Stat. 6. The next month, the White House disbanded its COVID–19 Response Team, which was responsible for many of the challenged communications in this case. Enjoining the Government defendants, therefore, is unlikely to affect the ������ٴڴǰ�����’ content-moderation decisions.[11]

C

We conclude briefly with the plaintiffs’ “right to listen” theory. The individual plaintiffs claim an interest in reading and engaging with the content of other speakers on social media. The First Amendment, they argue, protects that interest. Thus, the plaintiffs assert injuries based on the restrictions that countless other social-media users have experienced.

This theory is startlingly broad, as it would grant all social-media users the right to sue over someone else’s censorship — at least so long as they claim an interest in that person’s speech. This Court has “never accepted such a boundless theory of ���ٲ��Ի徱�Բ�.” Already, LLC v. Nike, Inc., , 99 (2013). While we have recognized a “ First Amendment right to ‘receive information and ideas,’ ” we have identified a cognizable injury only where the listener has a concrete, specific connection to the speaker. Kleindienst v. Mandel, 408 U.S. 753, 762 (1972). For instance, in Mandel, we agreed that a group of professors had a First Amendment interest in challenging the visa denial of a person they had invited to speak at a conference. Id., at 762–765. And in Virginia Bd. of Pharmacy v. Virginia Citizens Consumer Council, Inc., we concluded that prescription-drug consumers had an interest in challenging the prohibition on advertising the price of those drugs. 425 U.S. 748, 756–757 (1976).

Attempting to satisfy this requirement, the plaintiffs emphasize that hearing unfettered speech on social media is critical to their work as scientists, pundits, and activists. But they do not point to any specific instance of content moderation that caused them identifiable harm. They have therefore failed to establish an injury that is sufficiently “concrete and particularized.” Lujan, 504 U. S., at 560.

The state plaintiffs, claiming their own version of the “right to listen” theory, assert a sovereign interest in hearing from their citizens on social media. See 83 F. 4th, at 372–373. But this theory suffers from the same flaws as the individual plaintiffs’ theory. The States have not identified any specific speakers or topics that they have been unable to hear or follow.

The States cite this supposed sovereign injury as a basis for asserting third-party standing on behalf of “the citizens they would listen to.” Brief for Respondents 30. But “[t]his argument is a thinly veiled attempt to circumvent the limits on parens patriae ���ٲ��Ի徱�Բ�.” Brackeen, 599 U. S., at 295, n. 11. Namely, States do not have “ ‘standing as parens patriae to bring an action against the Federal Government.’ ” Id., at 295.

The States, like the individual plaintiffs, have failed to establish a likelihood of standing.

* �ă* �ă*

The plaintiffs, without any concrete link between their injuries and the defendants’ conduct, ask us to conduct a review of the years-long communications between dozens of federal officials, across different agencies, with different social-media platforms, about different topics. This Court’s standing doctrine prevents us from “exercis[ing such] general legal oversight” of the other branches of Government. TransUnion, 594 U. S., at 423–424. We therefore reverse the judgment of the Fifth Circuit and remand the case for further proceedings consistent with this opinion.

It is so ordered.

Notes

[1]  Since the events of this suit, Twitter has merged into X Corp. and is now known as X. Facebook is now known as Meta Platforms. For the sake of clarity, we will refer to these platforms as Twitter and Facebook, as they were known during the vast majority of the events underlying this suit.

[2]  The District Court also enjoined the National Institute of Allergy and Infectious Diseases (NIAID) and the State Department, along with their officials and employees. 680 F. Supp. 3d, at 700–701, 704–705. The Fifth Circuit removed these entities and individuals from the injunction, however, so they are not before us. Missouri v. Biden, 83 F. 4th 350, 391 (2023).

[3]  Because we do not reach the merits, we express no view as to whether the Fifth Circuit correctly articulated the standard for when the Government transforms private conduct into state action.

[4]  The Fifth Circuit relied on the District Court’s factual findings, many of which unfortunately appear to be clearly erroneous. The District Court found that the defendants and the platforms had an “efficient report-and-censor relationship.” Missouri v. Biden, 680 F. Supp. 3d 630, 715 (WD La. 2023). But much of its evidence is inapposite. For instance, the court says that Twitter set up a “streamlined process for censorship requests” after the White House “bombarded” it with such requests. Ibid., n. 662 (internal quotation marks omitted). The record it cites says nothing about “censorship requests.” See App. 639–642. Rather, in response to a White House official asking Twitter to remove an impersonation account of President Biden’s granddaughter, Twitter told the official about a portal that he could use to flag similar issues. Ibid. This has nothing to do with COVID–19 misinformation. The court also found that “[a] drastic increase in censorship . . . directly coincided with Defendants’ public calls for censorship and private demands for censorship.” 680 F. Supp. 3d, at 715. As to the “calls for censorship,” the court’s proof included statements from Members of Congress, who are not parties to this suit. Ibid., and n. 658. Some of the evidence of the “increase in censorship” reveals that Facebook worked with the CDC to update its list of removable false claims, but these examples do not suggest that the agency “demand[ed]” that it do so. Ibid. Finally, the court, echoing the plaintiffs’ proposed statement of facts, erroneously stated that Facebook agreed to censor content that did not violate its policies. Id., at 714, n. 655. Instead, on several occasions, Facebook explained that certain content did not qualify for removal under its policies but did qualify for other forms of moderation.

[5]  The Fifth Circuit held that States “sustain a direct injury when the social-media accounts of state officials are censored due to federal coercion.” 83 F. 4th, at 372. Because the State failed to show that its official was censored, we need not express a view on this theory.

[6]  Hines tries to link this restriction to the Surgeon General’s Office as well, suggesting that the White House and Surgeon General together pressured Facebook. But the record reveals that a White House official sent the relevant email, and Facebook responded only to White House officials. The Surgeon General’s Office was seemingly uninvolved. Thus, Hines cannot demonstrate that her past restriction is traceable to the Surgeon General’s Office. The plaintiffs do not attempt to draw any other connections between their restrictions and the Surgeon General’s Office.

[7]  The dissent does not dispute the Court’s assessment of these asserted links. Instead, the dissent draws links that Hines herself has not set forth, often based on injuries that Hines never claimed. Compare post, at 19–20, with Brief for Respondents 19–20; App. 628–632. For instance, the dissent says that in May 2021, Facebook began demoting content from accounts that repeatedly shared misinformation, purportedly due to White House pressure. Post, at 10, 19. Because Facebook frequently fact checked Hines’ posts, the dissent simply assumes (without citing Hines’ declarations) that her content was subsequently hidden from her friends’ feeds. Post, at 19. Likewise, pointing to an August 2021 policy change, the dissent concludes that the mid-July 2021 deplatforming of one of Hines’ groups rendered her other pages “non-recommendable.” Ibid. Hines, however, never claimed as much — and the plaintiffs bear the burden to establish standing by setting forth “specific facts.” Lujan v. Defenders of Wildlife, 504 U.S. 555, 561 (1992) (internal quotation marks omitted). It is especially important to hold the plaintiffs to their burden in a case like this one, where the record spans over 26,000 pages and the lower courts did not make any specific causation findings. As the Seventh Circuit has memorably put it, “[j]udges are not like pigs, hunting for truffles buried [in the record].” Gross v. Cicero, 619 F.3d 697, 702 (2010) (internal quotation marks omitted).

[8]  By acknowledging the real possibility that Facebook acted independently in suppressing Hines’ content, we are not applying a “new and heightened standard,” as the dissent claims. Post, at 20. The whole purpose of the traceability requirement is to ensure that “in fact, the asserted injury was the consequence of the defendants’ actions,” rather than of “the independent action” of a third party. Simon v. Eastern Ky. Welfare Rights Organization, , 42, 45 (1976). Nor is our analysis inconsistent with Department of Commerce v. New York, 588 U.S. 752 (2019). See post, at 19. There, the plaintiffs, including several States, challenged the Secretary of Commerce’s decision to reinstate a citizenship question on the census. 588 U. S., at 761, 764. They argued that this question would make noncitizens less likely to respond to the census, leading to an inaccurate population count and the concomitant loss of congressional seats and federal funding. Id., at 766–767. The plaintiffs’ injuries thus depended on the actions of third parties. Id., at 767–768. The District Court found that noncitizens had historically responded at lower rates than citizens to previous versions of the census (and other surveys) that included a citizenship question and that noncitizens were disproportionately likely to stop responding to those questionnaires once they reached the citizenship question. New York v. United States Dept. of Commerce, 351 F. Supp. 3d 502, 578–579 (SDNY 2019). Crediting those findings, this Court concluded that the plaintiffs “met their burden of showing that third parties will likely react in predictable ways to the citizenship question.” Department of Commerce, 588 U. S., at 768. The dissent suggests that it “would have been difficult for [the plaintiffs] to determine which noncitizen households failed to respond to the census because of a citizenship question and which had other reasons.” Post, at 20. But the evidence made clear that the citizenship question drove noncitizens’ lower response rates; the District Court made no findings about noncitizens’ response rates to the census generally. Here, by contrast, the evidence is murky. Facebook targeted Hines’ posts (and others like hers) before the White House entered the picture, meaning that Facebook had independent incentives to restrict Hines’ content. It is therefore difficult to say that the White House was responsible (even in part) for all of Hines’ later restrictions — especially absent clear links between White House content-moderation requests to Facebook and Facebook’s actions toward Hines. Cf. post, at 21.

[9]  According to a declaration submitted by the Surgeon General’s Chief of Staff, no one in that office met with the platforms to discuss their submissions “or otherwise had substantive communications with social media companies about the RFI.” 61 Record 19,480.

[10]  The dissent claims that the future injury prong is satisfied because Facebook continued to censor Hines at the time of her complaint and thereafter. Post, at 17. But the dissent gives short shrift to the key point: By the time Hines filed suit in August 2022, the White House was no longer engaged in any sort of “pressure campaign” toward Facebook. (Note that the dissent, in its 10-page recounting of the record, devotes only one paragraph to the events of 2022. Post, at 14.) Thus, when Hines sued, it was unlikely that Facebook’s actions were fairly traceable to the White House at the time — or would be going forward.

[11]  As with traceability, the dissent is wrong to claim that we are applying a “new and elevated standard for redressability.” Post, at 22. Far from holding plaintiffs to a “certainty” standard, ibid., we simply conclude that an injunction against the Government defendants is unlikely to stop the platforms from suppressing the plaintiffs’ speech. And while traceability and redressability are “ ‘often “flip sides of the same coin,” ’ ” post, at 22 (quoting FDA v. Alliance for Hippocratic Medicine, 602 U.S. 367, 380 (2024); emphasis added), that is not always the case. Facebook might continue to remove Hines’ posts under a policy that it adopted at the White House’s behest (thus satisfying traceability). But if the White House officials have already abandoned their pressure campaign, enjoining them is unlikely to prompt Facebook to stop enforcing the policy (thus failing redressability). Finally, by invoking Massachusetts v. EPA, it is the dissent that applies a new and loosened standard for redressability. Post, at 22. In that case, we explained that state plaintiffs are “entitled to special solicitude” when it comes to standing, and we conducted our analysis accordingly. 549 U.S. 497, 520 (2007). That “special solicitude” does not apply to Jill Hines, an individual.

SUPREME COURT OF THE UNITED STATES

_________________

No. 23–411

_________________

VIVEK H. MURTHY, SURGEON GENERAL, et al., PETITIONERS v. MISSOURI, et al.

on writ of certiorari to the united states court of appeals for the fifth circuit

[June 26, 2024]

Justice Alito, with whom Justice Thomas and Justice Gorsuch join, dissenting.

This case involves what the District Court termed “a far-reaching and widespread censorship campaign” conducted by high-ranking federal officials against Americans who expressed certain disfavored views about COVID–19 on social media. Missouri v. Biden, 680 F. Supp. 3d 630, 729 (WD La. 2023). Victims of the campaign perceived by the lower courts brought this action to ensure that the Government did not continue to coerce social media platforms to suppress speech. Among these victims were two States, whose public health officials were hampered in their ability to share their expertise with state residents; distinguished professors of medicine at Stanford and Harvard; a professor of psychiatry at the University of California, Irvine School of Medicine; the owner and operator of a news website; and Jill Hines, the director of a consumer and human rights advocacy organization. All these victims simply wanted to speak out on a question of the utmost public importance.

To protect their right to do so, the District Court issued a preliminary injunction, App. 278–285, and the Court of Appeals found ample evidence to support injunctive relief. See Missouri v. Biden, 83 F. 4th 350 (CA5 2023).

If the lower courts’ assessment of the voluminous record is correct, this is one of the most important free speech cases to reach this Court in years. Freedom of speech serves many valuable purposes, but its most important role is protection of speech that is essential to democratic self-government, see Snyder v. Phelps, 562 U.S. 443, 451–452 (2011), and speech that advances humanity’s store of knowledge, thought, and expression in fields such as science, medicine, history, the social sciences, philosophy, and the arts, see United States v. Alvarez, 567 U.S. 709, 751 (2012) (Alito, J., dissenting).

The speech at issue falls squarely into those categories. It concerns the COVID–19 virus, which has killed more than a million Americans.[1] Our country’s response to the COVID–19 pandemic was and remains a matter of enormous medical, social, political, geopolitical, and economic importance, and our dedication to a free marketplace of ideas demands that dissenting views on such matters be allowed. I assume that a fair portion of what social media users had to say about COVID–19 and the pandemic was of little lasting value. Some was undoubtedly untrue or misleading, and some may have been downright dangerous. But we now know that valuable speech was also suppressed.[2] That is what inevitably happens when entry to the marketplace of ideas is restricted.

Of course, purely private entities like newspapers are not subject to the First Amendment, and as a result, they may publish or decline to publish whatever they wish. But government officials may not coerce private entities to suppress speech, see National Rifle Association of America v. Vullo, 602 U.S. 175 (2024), and that is what happened in this case.

The record before us is vast. It contains evidence of communications between many different government actors and a variety of internet platforms, as well as evidence regarding the effects of those interactions on the seven different plaintiffs. For present purposes, however, I will focus on (a) just a few federal officials (namely, those who worked either in the White House or the Surgeon General’s office), (b) only one of the most influential social media platforms, Facebook, and (c) just one plaintiff, Jill Hines, because if any of the plaintiffs has standing, we are obligated to reach the merits of this case. See Rumsfeld v. Forum for Academic and Institutional Rights, Inc., 547 U.S. 47, 52, n. 2 (2006).

With the inquiry focused in this way, here is what the record plainly shows. For months in 2021 and 2022, a coterie of officials at the highest levels of the Federal Government continuously harried and implicitly threatened Facebook with potentially crippling consequences if it did not comply with their wishes about the suppression of certain COVID–19-related speech. Not surprisingly, Facebook repeatedly yielded. As a result Hines was indisputably injured, and due to the officials’ continuing efforts, she was threatened with more of the same when she brought suit. These past and threatened future injuries were caused by and traceable to censorship that the officials coerced, and the injunctive relief she sought was an available and suitable remedy. This evidence was more than sufficient to establish Hines’s standing to sue, see Lujan v. Defenders of Wildlife, 504 U.S. 555, 561–562 (1992), and consequently, we are obligated to tackle the free speech issue that the case presents. The Court, however, shirks that duty and thus permits the successful campaign of coercion in this case to stand as an attractive model for future officials who want to control what the people say, hear, and think.

That is regrettable. What the officials did in this case was more subtle than the ham-handed censorship found to be unconstitutional in Vullo, but it was no less coercive. And because of the perpetrators’ high positions, it was even more dangerous. It was blatantly unconstitutional, and the country may come to regret the Court’s failure to say so. Officials who read today’s decision together with Vullo will get the message. If a coercive campaign is carried out with enough sophistication, it may get by. That is not a message this Court should send.

In the next section of this opinion, I will recount in some detail what was done by the officials in this case, but in considering the coercive impact of their conduct, two prominent facts must be kept in mind.

First, social media have become a leading source of news for many Americans,[3] and with the decline of other media, their importance may grow.

Second, internet platforms, although rich and powerful, are at the same time far more vulnerable to Government pressure than other news sources. If a President dislikes a particular newspaper, he (fortunately) lacks the ability to put the paper out of business. But for Facebook and many other social media platforms, the situation is fundamentally different. They are critically dependent on the protection provided by §230 of the Communications Decency Act of 1996, 47 U. S. C. §230, which shields them from civil liability for content they spread. They are vulnerable to antitrust actions; indeed, Facebook CEO Mark Zuckerberg has described a potential antitrust lawsuit as an “existential” threat to his company.[4] And because their substantial overseas operations may be subjected to tough regulation in the European Union and other foreign jurisdictions, they rely on the Federal Government’s diplomatic efforts to protect their interests.

For these and other reasons,[5] internet platforms have a powerful incentive to please important federal officials, and the record in this case shows that high-ranking officials skillfully exploited Facebook’s vulnerability. When Facebook did not heed their requests as quickly or as fully as the officials wanted, the platform was publicly accused of “killing people” and subtly threatened with retaliation.

Not surprisingly these efforts bore fruit. Facebook adopted new rules that better conformed to the officials’ wishes, and many users who expressed disapproved views about the pandemic or COVID–19 vaccines were “deplatformed” or otherwise injured.

I

A

I begin by recounting the White House-led campaign to coerce Facebook. The story starts in early 2021, when White House officials began communicating with Facebook about the spread of misinformation about COVID–19 on its platform. Their emails started as questions, e.g., “Can you also give us a sense of misinformation that might be falling outside of your removal polices?” 10 Record 3397. But when the White House did not get the results it wanted, its questions quickly turned to virtual demands. And sometimes, those statements were paired with explicit references to potential consequences.

We may begin this account with an exchange that occurred in March 2021, when the Washington Post reported that Facebook was conducting a study that examined whether posts on the platform questioning COVID–19’s severity or the vaccines’ efficacy dissuaded some Americans from being vaccinated.[6] The study noted that Facebook’s rules permitted some of this content to circulate. Rob Flaherty, the White House Director of Digital Strategy, promptly emailed Facebook about the report. The subject line of his email contained this accusation: “You are hiding the ball.” 30 id., at 9366. Flaherty noted that the White House was “gravely concerned that [Facebook] is one of the top drivers of vaccine hesitancy,” and he demanded to know how Facebook was trying to solve the problem. Id., at 9365. In his words, “we want to know that you’re trying, we want to know how we can help, and we want to know that you’re not playing a shell game with us when we ask you what is going on.” Ibid.

Andy Slavitt, the White House Senior Advisor for the COVID–19 Response, chimed in with similar complaints. “[R]elative to othe[r]” platforms, he said, “interactions with Facebook are not straightforward” even though the misinformation problems there, in his view, were “worse.” Id., at 9364. According to Slavitt, the White House did not believe that Facebook was “trying to solve the problem,” so he informed Facebook that “[i]nternally we have been considering our options on what to do about it.” Ibid.

Facebook responded apologetically to this and other missives. It acknowledged that “[w]e obviously have work to do to gain your trust.” Id., at 9365. And after a follow-up conversation, the platform promised Flaherty and Slavitt that it would adopt additional policies to “reduc[e] virality of vaccine hesitancy content.” Id., at 9369. In particular, Facebook promised to “remove [any] Groups, Pages, and Accounts” that “disproportionately promot[e] . . . sensationalized content” about the risks of vaccines, even though it acknowledged that user stories about their experiences and those of family members or friends were “ofte[n] true.” Ibid. Facebook also promised to share additional data with the White House, ibid., but Flaherty was not fully satisfied. He said that the additional data Facebook offered was not “going to get us the info we’re looking for,” but “it shows to me that you at least understand the ask.” Id., at 9368.

In April, Flaherty again demanded information on the “actions and changes” Facebook was taking “to ensure you’re not making our country’s vaccine hesitancy problem worse.” Id., at 9371. To emphasize his urgency, Flaherty likened COVID–19 misinformation to misinformation that led to the January 6 attack on the Capitol. Ibid. Facebook, he charged, had helped to “increase skepticism” of the 2020 election, and he claimed that “an insurrection . . . was plotted, in large part, on your platform.” Ibid. He added: “I want some assurances, based in data, that you are not doing the same thing again here.” Ibid. Facebook was surprised by these remarks because it “thought we were doing a better job” communicating with the White House, but it promised to “more clearly respon[d]” in the future. Ibid.

The next week, Facebook officers spoke with Slavitt and Flaherty about reports of a rare blood clot caused by the Johnson & Johnson vaccine. Id., at 9385. The conversation quickly shifted when the White House noticed that one of the most-viewed vaccine-related posts from the past week was a Tucker Carlson video questioning the efficacy of the Johnson & Johnson vaccine. Id., at 9376, 9388. Facebook informed the White House that the video did not “qualify for removal under our policies” and thus would be demoted instead, ibid., but that answer did not please Flaherty. “How was this not violative?” he queried, and “[w]hat exactly is the rule for removal vs demoting?” Id., at 9387. Then, for the second time in a week, he invoked the January 6 attack: “Not for nothing, but last time we did this dance, it ended in an insurrection.” Id., at 9388. When Facebook did not respond promptly, he made his demand more explicit: “These questions weren’t rhetorical.” Id., at 9387.

If repeated accusations that Facebook aided an insurrection did not sufficiently convey the White House’s displeasure, Flaherty and Slavitt made sure to do so by phone.[7] In one call, both officials chided Facebook for not being “straightforward” and not “play[ing] ball.” Committee Report 141–142. Flaherty also informed Facebook that he was reporting on the COVID–19 misinformation problem to the President. Id., at 136.

After a second call, a high-ranking Facebook executive perceived that Slavitt was “outraged — not too strong a word to describe his reaction” — that the platform had not removed a fast-spreading meme suggesting that the vaccines might cause harm. Id., at 295. The executive had “countered that removing content like that would represent a significant incursion into traditional boundaries of free expression in the US,” but Slavitt was unmoved, in part because he presumed that other platforms “would never accept something like this.” Ibid.

A few weeks later, White House Press Secretary Jen Psaki was asked at a press conference about Facebook’s decision to keep former President Donald Trump off the platform. See Press Briefing by Press Secretary Jen Psaki and Secretary of Agriculture Tom Vilsack (May 5, 2021) (hereinafter May 5 Press Briefing).[8] Psaki deflected that question but took the opportunity to call on platforms like Facebook to “ ‘stop amplifying untrustworthy content . . . , especially related to COVID–19, vaccinations, and elections.’ ” 78 Record 25170. In the same breath, Psaki reminded the platforms that President Biden “ ‘supports . . . a robust anti-trust program.’ ” Id., at 25171 (emphasis deleted); May 5 Press Briefing.

Around this same time, Flaherty and Slavitt were in- terrogating Facebook on the mechanics of its content- moderation rules for COVID–19 misinformation. 30 Record 9391, 9397. Flaherty also forwarded to Facebook a “COVID–19 Vaccine Misinformation Brief ” that had been drafted by outside researchers and was “informing thinking” in the White House on what Facebook’s policies should be. 52 id., at 16186. This document recommended that Facebook strengthen its efforts against misinformation in several ways. It recommended the adoption of “progressively severe penalties” for accounts that repeatedly posted misinformation, and it proposed that Facebook make it harder for users to find “anti-vaccine or vaccine-hesitant propaganda” from other users. Ibid. Facebook declined to adopt some of these suggestions immediately, but it did “se[t] up more dedicated monitoring for [COVID] vaccine content” and adopted a policy of “stronger demotions [for] a broader set of content.” 30 id., at 9396.

The White House responded with more questions. Acknowledging that he sounded “like a broken record,” Flaherty interrogated Facebook about “how much content is being demoted, and how effective [Facebook was] at mitigating reach, and how quickly.” Id., at 9395. Later, Flaherty chastised Facebook for failing to prevent some vaccine-hesitant content from showing up through the ������ٴڴǰ���’s search function. Id., at 9400. “ ‘[R]emoving bad information from search’ is one of the easy, low-bar things you guys do to make people like me think you’re taking action,” he said. Id., at 9399. “If you’re not getting that right, it raises even more questions about the higher bar stuff.” Ibid. A few weeks after this latest round of haranguing, Facebook expanded penalties for individual Facebook accounts that repeatedly shared content that fact-checkers deemed misinformation; henceforth, all of those individuals’ posts would show up less frequently in their friends’ news feeds. See 9 id., at 2697; Facebook, Taking Action Against People Who Repeatedly Share Misinformation (May 26, 2021).[9]

Perhaps the most intense period of White House pressure began a short time later. On July 15, Surgeon General Vivek Murthy released an advisory titled “Confronting Health Misinformation.” 78 Record 25171, 25173. Dr. Murthy suggested, among other things, algorithmic changes to demote misinformation and additional consequences for misinformation “ ‘super-spreaders.’ ” U. S. Public Health Service, Confronting Health Misinformation: The U. S. Surgeon General’s Advisory on Building a Healthy Information Environment 12 (2021).[10] Dr. Murthy also joined Psaki at a press conference, where he asked the platforms to take “much, much more . . . aggressive action” to combat COVID–19 misinformation “because it’s costing people their lives.” Press Briefing by Press Secretary Jen Psaki and Surgeon General Dr. Vivek H. Murthy (July 15, 2021).[11]

At the same press conference, Psaki singled out Facebook as a primary driver of misinformation and asked the platform to make several changes. Facebook “should provide, publicly and transparently, data on the reach of COVID–19 [and] COVID vaccine misinformation.” Ibid. It “needs to move more quickly to remove harmful, violative posts.” Ibid. And it should change its algorithm to promote “quality information sources.” Ibid. These recommendations echoed Slavitt’s and Flaherty’s private demands from the preceding months — as Psaki herself acknowledged. The White House “engage[s] with [Facebook] regularly,” she said, and Facebook “certainly understand[s] what our asks are.” Ibid. Apparently, the White House had not gotten everything it wanted from those private conversations, so it was turning up the heat in public.

Facebook responded by telling the press that it had partnered with the White House to counter misinformation and that it had “removed accounts that repeatedly break the rules” and “more than 18 million pieces of COVID misinformation.” 78 Record 25174. But at another press briefing the next day, Psaki said these efforts were “[c]learly not” sufficient and expressed confidence that Facebook would “make decisions about additional steps they can take.” See id., at 25175; Press Briefing by Press Secretary Jen Psaki (July 16, 2021).[12]

That same day, President Biden told reporters that social media platforms were “ ‘killing people’ ” by allowing COVID-related misinformation to circulate. 78 Record 25174, 25212. At oral argument, the Government suggested that the President later disclaimed any desire to hold the platforms accountable for misinformation, Tr. of Oral Arg. 34–35, but that is not so. The President’s so-called clarification, like many other statements by Government officials, called on “ ‘Facebook’ ” to “ ‘do something about the misinformation’ ” on its platform. B. Klein, M. Vazquez, & K. Collins, Biden Backs Away From His Claim That Facebook Is ‘Killing People’ by Allowing COVID Misinformation, CNN (July 19, 2021).[13]

And far from disclaiming potential regulatory action, the White House confirmed that it had not “ ‘taken any options off the table.’ ” Ibid. In fact, the day after the President’s supposed clarification, the White House Communications Director commended the President for “speak[ing] very aggressively” and affirmed that platforms “certainly . . . should be held accountable” for publishing misinformation. 61 Record 19400–19401. Indeed, she said that the White House was “reviewing” whether §230 should be amended to open the platforms to suit. Id., at 19400.

Facebook responded quickly. The same day the President made his “killing people” remark, the platform reached out to Dr. Murthy to determine “the scope of what the White House expects from us on misinformation going forward.” 9 id., at 2690. The next day, Facebook asked officials about how to “get back to a good place” with the White House. 30 id., at 9403. And soon after, Facebook sent an email saying that it “hear[d]” the officials’ “call for us to do more,” and promptly assured the White House that it would comply. 9 id., at 2706. In spite of the White House’s inflammatory rhetoric, Facebook at all times went out of its way to strike a conciliatory tone. Only two days after the President’s remark — and before his supposed clarification — Facebook assured Dr. Murthy that, though “it’s not great to be accused of killing people,” Facebook would “find a way to deescalate and work together collaboratively.” Id., at 2713.

Concrete changes followed in short order. In early August, the Surgeon General’s Office reached out to Facebook for “an update of any new/additional steps you are taking with respect to health misinformation in light of ” the July 15 advisory. Id., at 2703. In response, Facebook informed the Surgeon General that it would soon “expan[d] [its] COVID policies to further reduce the spread of potentially harmful content.” Id., at 2701.

White House-Facebook conversations about misinformation did not end there. In September, the Wall Street Journal wrote about the spread of misinformation on Facebook, and Facebook preemptively reached out to the White House to clarify. 8 id., at 2681. Flaherty asked (again) for information on “how big the problem is, what solutions you’re implementing, and how effective they’ve been.” Ibid.

Then in October, the Washington Post published yet another story suggesting that Facebook knew more than it let on about the spread of misinformation. Flaherty emailed the link to Facebook with the subject line: “not even sure what to say at this point.” Id., at 2676. And the Surgeon General’s Office indicated both publically and privately that it was disappointed in Facebook. See @Surgeon_General, X (Oct. 29, 2021) (accusing Facebook of “lacking . . . transparency and accountability”);[14] 9 Record 2708. Facebook offered to speak with both the White House and the Surgeon General’s Office to assuage concerns. 8 id., at 2676.

Interactions related to COVID–19 misinformation continued until at least June 2022. Id., at 2663. At that point, Facebook proposed discontinuing its reports on misinformation, but assured the White House that it would be “happy to continue, or to pick up at a later date, . . . if we hear from you that this continues to be of value.” Ibid. Flaherty asked Facebook to continue reporting on misinformation because the Government was preparing to roll out COVID–19 vaccines for children under five years old and, “[o]bviously,” that rollout “ha[d] the potential to be just as charged” as other vaccine-related controversies. Ibid. Flaherty added that he “[w]ould love to get a sense of what you all are planning here,” and Facebook agreed to provide information for as long as necessary. Ibid.

What these events show is that top federal officials continuously and persistently hectored Facebook to crack down on what the officials saw as unhelpful social media posts, including not only posts that they thought were false or misleading but also stories that they did not claim to be literally false but nevertheless wanted obscured. See, e.g., 30 id., at 9361, 9365, 9369, 9385–9388. And Facebook’s reactions to these efforts were not what one would expect from an independent news source or a journalistic entity dedicated to holding the Government accountable for its actions. Instead, Facebook’s responses resembled that of a subservient entity determined to stay in the good graces of a powerful taskmaster. Facebook told White House officials that it would “work . . . to gain your trust.” Id., at 9365. When criticized, Facebook representatives whimpered that they “thought we were doing a better job” but promised to do more going forward. Id., at 9371. They pleaded to know how they could “get back to a good place” with the White House. Id., at 9403. And when denounced as “killing people,” Facebook responded by expressing a desire to “work together collaboratively” with its accuser. 9 id., at 2713; 78 id., at 25174. The picture is clear.

B

While all this was going on, Jill Hines and others were subjected to censorship. Hines serves as the co-director of Health Freedom Louisiana, an organization that advocated against vaccine and mask mandates during the pandemic. Over the course of the pandemic — and while the White House was pressuring Facebook — the platform repeatedly censored Hines’s speech.

For instance, in the summer and fall of 2021, Facebook removed two groups that Hines had formed to discuss the vaccine. 4 id., at 1313–1315. In January 2022, Facebook restricted posts from Hines’s personal page “for 30 days . . . for sharing the image of a display board used in a legislative hearing that had Pfizer’s preclinical trial data on it.” Id., at 1313. In late May, Facebook restricted Hines for 90 days for sharing an article about “increased emergency calls for teens with myocarditis following [COVID] vaccination.” Id., at 1313–1314. Hines’s public pages, Reopen Louisiana and Health Freedom Louisiana, were subjected to similar treatment. Facebook’s disciplinary actions meant that both public pages suffered a drop in viewership; as Hines put it, “Each time you build viewership up [on a page], it is knocked back down with each violation.” Id., at 1314. And from February to April 2023, Facebook issued warnings and violations for several vaccine-related posts shared on Hines’s personal and public pages, including a post by Robert F. Kennedy, Jr., and an article entitled “ ‘Some Americans Shouldn’t Get Another COVID-19 Vaccine Shot, FDA Says.’ ” 78 id., at 25503–25506. The result was that “[n]o one else was permitted to view or engage with the[se] post[s].” Id., at 25503.

II

Hines and the other plaintiffs in this case brought this suit and asked for an injunction to stop the censorship campaign just described. To maintain that suit, they needed to show that they (1) were imminently threatened with an injury in fact (2) that is traceable to the defendants and (3) that could be redressed by the court. Lujan, 504 U. S., at 560–561; ��’S��� v. Littleton, , 496 (1974). Hines satisfied all these requirements.

A

Injury in fact. Because Hines sought and obtained a preliminary injunction, it was not enough for her to show that she had been injured in the past. Instead, she had to identify a “real and immediate threat of repeated injury” that existed at the time she sued — that is, on August 2, 2022. ��’S���, 414 U. S., at 496; see also Friends of the Earth, Inc. v. Laidlaw Environmental Services (TOC), Inc., 528 U.S. 167, 191 (2000); Mollan v. Torrance, 9 Wheat. 537, 539 (1824).

The Government concedes that Hines suffered past injury, but it claims that she did not make the showing needed to obtain prospective relief. See Brief for Petitioners 17. Both the District Court and the Court of Appeals rejected this argument and found that Hines had shown that she was likely to be censored in the future. 680 F. Supp. 3d, at 713; 83 F. 4th, at 368–369. We have previously examined such findings under the “clearly erroneous” test. See Duke Power Co. v. Carolina Environmental Study Group, Inc., 438 U.S. 59, 77 (1978). But no matter what test is applied, the record clearly shows that Hines was still being censored when she sued — and that the censorship continued thereafter. See supra, at 15–16. That was sufficient to establish the type of injury needed to obtain injunctive relief. ��’S���, 414 U. S., at 496; see also County of Riverside v. McLaughlin, 500 U.S. 44, 51 (1991).

B

Traceability. To sue the White House officials, Hines had to identify a “causal connection” between the actions of those officials and her censorship. Bennett v. Spear, 520 U.S. 154, 169 (1997). Hines did not need to prove that it was only because of those officials’ conduct that she was censored. Rather, as we held in Department of Commerce v. New York, 588 U.S. 752 (2019), it was enough for her to show that one predictable effect of the officials’ action was that Facebook would modify its censorship policies in a way that affected her. Id., at 768.

Hines easily met that test, and her traceability theory is at least as strong as the State of New York’s in the Department of Commerce case. There, the State claimed that it would be hurt by a census question about citizenship. The State predicted that the question would dissuade some noncitizen households from complying with their legal duty to complete the form, and it asserted that this in turn could cause the State to lose a seat in the House of Representatives, as well as federal funds that are distributed on the basis of population. Id., at 766–767. Although this theory depended on illegal conduct by third parties and an attenuated chain of causation, the Court found that the State had established traceability. It was enough, the Court held, that the failure of some aliens to respond to the census was “likely attributable” to the Government’s introduction of a citizenship question. Id., at 768.

This is not a demanding standard, and Hines made the requisite showing — with room to spare. Recall that officials from the White House and Surgeon General’s Office repeatedly hectored and implicitly threatened Facebook to suppress speech expressing the viewpoint that Hines espoused. See supra, at 6–15. Censorship of Hines was the “predictable effect” of these efforts. Department of Commerce, 588 U. S., at 768. Or, to put the point in different terms, Facebook would “likely react in predictable ways” to this unrelenting pressure. Ibid.

This alone was sufficient to show traceability, but here there is even more direct proof. On numerous occasions, the White House officials successfully pushed Facebook to tighten its censorship policies, see supra, at 7, 10, 13, and those policies had implications for Hines.[15] First, in March 2021, the White House pressured Facebook into implementing a policy of removing accounts that “disproportionately promot[e] . . . sensationalized content” about vaccines. Supra, at 7. Later that year, Facebook removed two of Hines’s groups, which posted about vaccines. Supra, at 15. And when Hines sued in August 2022, she reported that her personal page was “currently restricted” for sharing vaccine-related content and, thus, that she was “under constant threat of being completely deplatformed.” 4 Record 1314.

Second, in May, Facebook told Slavitt that it would “se[t] up more dedicated monitoring” of vaccine content and apply demotions to “a broader set of content.” Supra, at 10. Then, a few weeks later, Facebook also increased demotions of posts by individual Facebook accounts that repeatedly shared misinformation. Ibid. Hines says that she was repeatedly fact-checked for posting about the vaccines, see supra, at 15–16; 4 Record 1314, so these policy changes apparently increased the risk that posts from her personal account would have been hidden from her friends’ Facebook feeds.

Third, in response to the July 2021 comments from the White House and the Surgeon General, Facebook made more changes. Supra, at 13. And from the details Hines provides about her posting history, this policy change would have affected her. For one thing, Facebook “rendered ‘non-recommendable’ ” any page linked to another account that had been “removed” for spreading misinformation about COVID–19. 9 Record 2701. Hines says that two of her groups were removed for alleged COVID misinformation around this time. Supra, at 15; 4 Record 1315. So under the new policy, her other pages would apparently be non-recommendable. Perhaps for this reason, though Hines attempted to convince members of her deplatformed group to migrate to a substitute group, only about a quarter of its membership made the move before the substitute group too was removed. Ibid.

For another, Facebook “increas[ed] the strength of [its] demotions for COVID and vaccine-related content that third party fact checkers rate[d] as ‘Partly False’ or ‘Missing Context.’ ” 9 id., at 2701. And Facebook “ma[de] it easier to have Pages/Groups/Accounts demoted for sharing COVID and vaccine-related misinformation by . . . counting content removals” under Facebook’s COVID–19 policies “towards their demotion threshold.” Ibid. Under this new policy, Facebook would now consider Hines’s “numerous” community standards violations, 4 id., at 1314, when determining whether to make her posts less accessible to other users. So, for instance, when Hines received several citations in early 2023, this amendment would have governed Facebook’s decision to “downgrad[e] the visibility of [her] posts in Facebook’s News Feed (thereby limiting its reach to other users).” 78 id., at 25503. The record here amply shows traceability.

The Court reaches the opposite conclusion by applying a new and heightened standard. The Court notes that Facebook began censoring COVID–19-related misinformation before officials from the White House and the Surgeon General’s Office got involved. Ante, at 20; see also Brief for Petitioners 18. And in the Court’s view, that fact makes it difficult to untangle Government-caused censorship from censorship that Facebook might have undertaken anyway. See ante, at 20. That may be so, but in the Department of Commerce census case, it also would have been difficult for New York to determine which noncitizen households failed to respond to the census because of a citizenship question and which had other reasons. Nevertheless, the Court did not require New York to perform that essentially impossible operation because it was clear that a citizenship question would dissuade at least some noncitizen households from responding. As we explained, “Article III ‘requires no more than de facto causality,’ ” so a showing that a citizenship question affected some aliens sufficed. Department of Commerce, 588 U. S., at 768.

Here, it is reasonable to infer (indeed, the inference leaps out from the record) that the efforts of the federal officials affected at least some of Facebook’s decisions to censor Hines. All of Facebook’s demotion, content-removal, and deplatforming decisions are governed by its policies.[16] So when the White House pressured Facebook to amend some of the policies related to speech in which Hines engaged, those amendments necessarily impacted some of Facebook’s censorship decisions. Nothing more is needed. What the Court seems to want are a series of ironclad links — from a particular coercive communication to a particular change in Facebook’s rules or practice and then to a particular adverse action against Hines. No such chain was required in the Department of Commerce case, and neither should one be demanded here.

In addition to this heightened linkage requirement, the Court argues that Hines lacks standing because the threat of future injury dissipated at some point during summer 2022 when the officials’ pressure campaign tapered off. Ante, at 25, n. 10. But this argument errs in two critical respects. First, the effects of the changes the officials coerced persisted. Those changes controlled censorship decisions before and after Hines sued.

Second, the White House threats did not come with expiration dates, and it would be silly to assume that the threats lost their force merely because White House officials opted not to renew them on a regular basis. Indeed, the record suggests that Facebook did not feel free to chart its own course when Hines sued; rather, the platform had promised to continue reporting to the White House and remain responsive to its concerns for as long as the officials requested. Supra, at 14.

In short, when Hines sued in August 2022, there was still a link between the White House and the injuries she was presently suffering and could reasonably expect to suffer in the future. That is enough for traceability.

C

Redressability. Finally, Hines was required to show that the threat of future injury she faced when the complaint was filed “likely would be redressed” by injunctive relief. FDA v. Alliance for Hippocratic Medicine, 602 U.S. 367, 380 (2024). This required proof that a preliminary injunction would reduce Hines’s “risk of [future] harm . . . to some extent.” Massachusetts v. EPA, 549 U.S. 497, 526 (2007) (emphasis added). And as we recently explained, “[t]he second and third standing requirements — causation and redressability — are often ‘flip sides of the same coin.’ ” Alliance for Hippocratic Medicine, 602 U. S., at 380. Therefore, “[i]f a defendant’s action causes an injury, enjoining the action or awarding damages for the action will typically redress that injury.” Id., at 381.

Hines easily satisfied that requirement. For the reasons just explained, there is ample proof that Hines’s past injuries were a “predictable effect” of the Government’s censorship campaign, and the preliminary injunction was likely to prevent the continuation of the harm to at least “some extent.” Massachusetts v. EPA, 549 U. S., at 526.

The Court disagrees because Facebook “remain[s] free to enforce . . . even those [policies] tainted by initial governmental coercion.” Ante, at 26. But as with traceability, the Court applies a new and elevated standard for redressability, which has never required plaintiffs to be “certain” that a court order would prevent future harm. Larson v. Valente, 456 U.S. 228, 243–244, n. 15 (1982). In Massachusetts v. EPA, for example, no one could say that the relief sought — reconsideration by the EPA of its decision not to regulate the emission of greenhouse gases — would actually remedy the Commonwealth’s alleged injuries, such as the loss of land due to rising sea levels. The Court’s decision did not prevent the EPA from adhering to its prior decision, 549 U. S., at 534–535, and there was no way to know with any degree of certainty that any greenhouse gas regulations that the EPA might eventually issue would prevent the oceans from rising. Yet the Court found that the redressability requirement was met.

Similarly, in Department of Commerce, no one could say with any certainty that our decision barring a censorship question from the 2020 census questionnaire would prevent New York from losing a seat in the House of Representatives, 588 U. S., at 767, and in fact that result occurred despite our decision. S. Goldmacher, New York Loses House Seat After Coming Up 89 People Short on Census, N. Y. Times, Apr. 26, 2021.[17]

As we recently proclaimed in FDA v. Alliance for Hippocratic Medicine, Article III standing is an important component of our Constitution’s structural design. See 602 U. S., at 378–380. That doctrine is cheapened when the rules are not evenhandedly applied.

* �ă* �ă*

Hines showed that, when she sued, Facebook was censoring her COVID-related posts and groups. And because the White House prompted Facebook to amend its censorship policies, Hines’s censorship was, at least in part, caused by the White House and could be redressed by an injunction against the continuation of that conduct. For these reasons, Hines met all the requirements for Article III standing.

III

I proceed now to the merits of Hines’s First Amendment claim.[18] Government efforts to “dictat[e] the subjects about which persons may speak,” First Nat. Bank of Boston v. Bellotti, 435 U.S. 765, 784–785 (1978), or to suppress protected speech are “ ‘presumptively unconstitutional,’ ” Rosenberger v. Rector and Visitors of Univ. of Va., 515 U.S. 819, 830 (1995). And that is so regardless of whether the Government carries out the censorship itself or uses a third party “ ‘to accomplish what . . . is constitutionally forbidden.’ ” Norwood v. Harrison, 413 U.S. 455, 465 (1973).

As the Court held more than 60 years ago in Bantam Books, Inc. v. Sullivan, 372 U.S. 58 (1963), the Government may not coerce or intimidate a third-party intermediary into suppressing someone else’s speech. Id., at 67. Earlier this Term, we reaffirmed that important principle in National Rifle Association v. Vullo, 602 U. S., at 187–191. As we said there, “a government official cannot do indirectly what she is barred from doing directly,” id., at 190, and while an official may forcefully attempt to persuade, “[w]hat she cannot do . . . is use the power of the State to punish or suppress disfavored expression,” id., at 188.

In Vullo, the alleged conduct was blunt. The head of the state commission with regulatory authority over insurance companies allegedly told executives at Lloyd’s directly and in no uncertain terms that she would be “ ‘less interested’ ” in punishing the company’s regulatory infractions if it ceased doing business with the National Rifle Association. Id., at 183. The federal officials’ conduct here was more subtle and sophisticated. The message was delivered piecemeal by various officials over a period of time in the form of aggressive questions, complaints, insistent requests, demands, and thinly veiled threats of potentially fatal reprisals. But the message was unmistakable, and it was duly received.

The principle recognized in Bantam Books and Vullo requires a court to distinguish between permissible persuasion and unconstitutional coercion, and in Vullo, we looked to three leading factors that are helpful in making that determination: (1) the authority of the government officials who are alleged to have engaged in coercion, (2) the nature of statements made by those officials, and (3) the reactions of the third party alleged to have been coerced. 602 U. S., at 189–190, and n. 4, 191–194. In this case, all three factors point to coercion.

A

I begin with the authority of the relevant officials — high-ranking White House officials and the Surgeon General. High-ranking White House officials presumably speak for and may have the ability to influence the President, and as discussed earlier, a Presidential administration has the power to inflict potentially fatal damage to social media platforms like Facebook. See supra, at 5. Facebook appreciates what the White House could do, and President Biden has spoken openly about that power — as he has every right to do. For instance, he has declared that the “policy of [his] Administration [is] to enforce the antitrust laws to meet the challenges posed by . . . the rise of the dominant Internet platforms,” and he has directed the Attorney General and other agency heads to “enforce the antitrust laws . . . vigorously.” Promoting Competition in the American Economy, Executive Order No. 14036, 3 CFR 609 (2021).[19] He has also floated the idea of amending or repealing §230 of the Communications Decency Act. See, e.g., B. Klein, White House Reviewing Section 230 Amid Efforts To Push Social Media Giants To Crack Down on Misinformation, CNN (July 20, 2021) [20]; R. Kern, White House Renews Call To ‘Remove’ Section 230 Liability Shield, Politico (Sept. 8, 2022).[21]

Previous administrations have also wielded significant power over Facebook. In a data-privacy case brought jointly by the Department of Justice and the Federal Trade Commission, Facebook was required “to pay an unprecedented $5 billion civil penalty,” which is “among the largest civil penalties ever obtained by the federal government.” Press Release, Dept. of Justice, Facebook Agrees To Pay $5 Billion and Implement Robust New Protections of User Information in Settlement of Data-Privacy Claims (July 24, 2019).[22]

A matter that may well have been prominent in Facebook’s thinking during the period in question in this case was a dispute between the United States and the European Union over international data transfers. In 2020, the Court of Justice of the European Union invalidated the mechanism for transferring data between the European Union and United States because it did not sufficiently protect EU citizens from Federal Government surveillance. Data Protection Comm’r v. Facebook Ireland Limited, Case C–311/18 (2020). The EU-U. S. conflict over data privacy hindered Facebook’s international operations, but Facebook could not “resolve [the conflict] on its own.” N. Clegg & J. Newstead, Our Response to the Decision on Facebook’s EU-US Data Transfers, Meta (May 22, 2023).[23] Rather, the platform relied on the White House to negotiate an agreement that would preserve its ability to maintain its trans-Atlantic operations. K. Mackrael, EU Approves Data-Transfer Deal With U. S., Averting Potential Halt in Flows, Wall Street Journal, July 10, 2023.[24]

It is therefore beyond any serious dispute that the top-ranking White House officials and the Surgeon General possessed the authority to exert enormous coercive pressure.

B

1

Second, I turn to of the officials’ communications with Facebook, which possess all the hallmarks of coercion that we identified in Bantam Books and Vullo. Many of the White House’s emails were “phrased virtually as orders,” Bantam Books, 372 U. S., at 68, and the officials’ frequent follow-ups ensured that they were understood as such, id., at 63. To take a few examples, after Flaherty read an article about content causing vaccine hesitancy, he demanded “to know that [Facebook was] trying” to combat the issue and “to know that you’re not playing a shell game with us when we ask you what is going on.” 30 Record 9365; see supra, at 7. The next month, he requested “assurances, based in data,” that Facebook was not “making our country’s vaccine hesitancy problem worse.” 30 Record 9371; see supra, at 7–8. A week after that, he questioned Facebook about its policies “for removal vs demoting,” and when the platform did not promptly respond, he added: “These questions weren’t rhetorical.” 30 Record 9387; see supra, at 8. When Facebook provided the White House with some data it asked for, Flaherty thanked Facebook for demonstrating “that you at least understand the ask.” 30 Record 9368; see supra, at 7.

Various comments during the July pressure campaign likewise reveal that the White House and the Surgeon General’s Office expected compliance. At the press conference announcing the Surgeon General’s recommendations related to misinformation, Psaki noted that the White House “engage[s] with [Facebook] regularly,” and Facebook “certainly understand[s] what our asks are.” Supra, at 11. The next day, she expressed confidence that Facebook would “make decisions about additional steps they can take.” 78 Record 25175; see supra, at 12. And eventually, the Surgeon General’s Office prompted Facebook for “an update of any new/additional steps you are taking with respect to health misinformation in light of ” the July 15 advisory. 9 Record 2703; see supra, at 13.

These demands were coupled with “thinly veiled threats” of legal consequences. Bantam Books, 372 U. S., at 68. Three instances stand out. Early on, when the White House first expressed skepticism that Facebook was effectively combatting misinformation, Slavitt informed the platform that the White House was “considering our options on what to do about it.” 30 Record 9364; see supra, at 7. In other words, if Facebook did not “solve” its “misinformation” problem, the White House might unsheathe its potent authority. 30 Record 9364.

The threat was made more explicit in May, when Psaki paired a request for platforms to “ ‘stop amplifying untrustworthy content’ ” with a reminder that President Biden “ ‘supports . . . a robust anti-trust program.’ ” 78 id., at 25170–25171 (emphasis deleted); May 5 Press Briefing; see also supra, at 9. The Government casts this reference to legal consequences as a defense of individual Americans against censorship by the platforms. See Reply Brief 9. But Psaki’s full answer undermines that interpretation. Immediately after noting President Biden’s support for antitrust enforcement, Psaki added, “So his view is that there’s more that needs to be done to ensure that this type of . . . life-threatening information is not going out to the American public.” May 5 Press Briefing. The natural interpretation is that the White House might retaliate if the platforms allowed free speech, not if they suppressed it.

Finally, in July, the White House asserted that the platforms “should be held accountable” for publishing misinformation. 61 Record 19400; see supra, at 11–13. The totality of this record — constant haranguing, dozens of demands for compliance, and references to potential consequences — evince “a scheme of state censorship.” Bantam Books, 372 U. S., at 72.

2

The Government tries to spin these interactions as fairly benign. In its telling, Flaherty, Slavitt, and other officials merely “asked the platforms for information” and then “publicly and privately criticized the platforms for what the officials perceived as a . . . failure to live up to the ������ٴڴǰ�����’ commitments.” Brief for Petitioners 31. References to consequences, the Government claims, were “fleeting and general” and “cannot plausibly be characterized as coercive threats.” Id., at 32.

This characterization is not true to what happened. Slavitt and Flaherty did not simply ask Facebook for information. They browbeat the platform for months and made it clear that if it did not do more to combat what they saw as misinformation, it might be called to account for its shortcomings. And as for the supposedly “fleeting” nature of the numerous references to potential consequences, death threats can be very effective even if they are not delivered every day.

The Government also defends the officials’ actions on the ground that “[t]he President and his senior aides are entitled to speak out on such matters of pressing public concern.” Reply Brief 11. According to the Government, the officials were simply using the President’s “bully pulpit” to “inform, persuade, and protect the public.” Brief for Petitioners 5, 24.

This argument introduces a new understanding of the term “bully pulpit,” which was coined by President Theodore Roosevelt to denote a President’s excellent (i.e., “bully” [25]) position (i.e., his “pulpit”) to persuade the public.[26] But Flaherty, Slavitt, and other officials who emailed and telephoned Facebook were not speaking to the public from a figurative pulpit. On the contrary, they were engaged in a covert scheme of censorship that came to light only after the plaintiffs demanded their emails in discovery and a congressional Committee obtained them by subpoena. See Committee Report 1–2. If these communications represented the exercise of the bully pulpit, then everything that top federal officials say behind closed doors to any private citizen must also represent the exercise of the President’s bully pulpit. That stretches the concept beyond the breaking point.

In any event, the Government is hard-pressed to find any prior example of the use of the bully pulpit to threaten censorship of private speech. The Government cites four instances in which past Presidents commented publicly about the performance of the media. President Reagan lauded the media for “tough reporting” on drugs. Reagan Presidential Library & Museum, Remarks to Media Executives at a White House Briefing on Drug Abuse (Mar. 7, 1988).[27] But he never threatened to do anything to media outlets that were soft on the issue of drugs. President Theodore Roosevelt “lambasted ‘muck-raking’ journalists” as “ ‘one of the most potent forces for evil’ ” and encouraged journalists to speak truth, rather than slander. Brief for Petitioners 24 (quoting The American Presidency Project, Remarks at the Laying of the Cornerstone of the Office Building of the House of Representatives (Apr. 14, 1906)).[28] But his comment did not threaten any action against the muckrakers, see Goodwin 480–487, and it is unclear what he could have done to them. President George W. Bush denounced pornography as “debilitating” for “communities, marriages, families, and children.” Presidential Proclamation No. 7725, 3 CFR 129 (2003 Comp.). But he never threatened to take action against pornography that was not “obscene” within the meaning of our precedents.

The Government’s last example is a 1915 speech in which President Wilson deplored false reporting that the Japanese were using Turtle Bay, California, as a naval base. The American Presidency Project, Address at the Associated Press Luncheon in New York City (Apr. 20, 1915).[29] Speaking to a gathering of reporters, President Wilson proclaimed: “We ought not to permit that sort of thing to use up the electrical energy of the [telegraph] wires, because its energy is malign, its energy is not of the truth, its energy is mischief.” Ibid. Wilson’s comment is best understood as metaphorical and hortatory, not as a legal threat. And in any event, it is hard to see how he could have brought about censorship of telegraph companies because the Mann-Elkins Act, enacted in 1910, deemed them to be common carriers, and that meant that they were obligated to transmit all messages regardless of content. See 36Stat. 544–545; T. Wu, A Brief History of American Telecommunications Regulation, in 5 Oxford International Encyclopedia of Legal History 95 (2007). Thus, none of these examples justifies the conduct at issue here.

C

Finally, Facebook’s responses to the officials’ persistent inquiries, criticisms, and threats show that the platform perceived the statements as something more than mere recommendations. Time and time again, Facebook responded to an angry White House with a promise to do better in the future. In March, Facebook attempted to assuage the White House by acknowledging “[w]e obviously have work to do to gain your trust.” 30 Record 9365. In April, Facebook promised to “more clearly respon[d] to [White House] questions.” Id., at 9371. In May, Facebook “committed to addressing the defensive work around misinformation that you’ve called on us to address.” 9 id., at 2698. In July, Facebook reached out to the Surgeon General after “the President’s remarks about us” and emphasized its efforts “to better understand the scope of what the White House expects from us on misinformation going forward.” Id., at 2690. And of course, as we have seen, Facebook repeatedly changed its policies to better address the White House’s concerns. See supra, at 7, 10, 13.

The Government’s primary response is that Facebook occasionally declined to take its suggestions. Reply Brief 11; see, e.g., supra, at 10. The implication is that Facebook must have chosen to undertake all of its anti-misinformation efforts entirely of its own accord.

That is bad logic, and in any event, the record shows otherwise. It is true that Facebook voluntarily undertook some anti-misinformation efforts and that it declined to make some requested policy changes. But the interactions recounted above unmistakably show that the White House was insistent that Facebook should do more than it was doing on its own, see, e.g., supra, at 11–12, and Facebook repeatedly yielded — even if it did not always give the White House everything it wanted.

Internal Facebook emails paint a clear picture of subservience. The platform quickly realized that its “handling of [COVID] misinformation” was “importan[t]” to the White House, so it looked for ways “to be viewed as a trusted, transparent partner” and “avoid . . . public spat[s].” Committee Report 181, 184, 188. After the White House blamed Facebook for aiding an insurrection, the platform realized that it was at a “crossroads . . . with the White House.” Id., at 294. “Given what is at stake here,” one Facebook employee proposed reevaluating the company’s “internal methods” to “see what further steps we may/may not be able to take.” Id., at 295. This reevaluation led to one of Facebook’s policy changes. See supra, at 8–10.

Facebook again took stock of its relationship with the White House after the President’s accusation that it was “killing people.” Internally, Facebook saw little merit in many of the White House’s critiques. One employee labeled the White House’s understanding of misinformation “completely unclear” and speculated that “it’s convenient for them to blame us” “when the vaccination campaign isn’t going as hoped.” Committee Report 473. Nonetheless, Facebook figured that its “current course” of “in effect explaining ourselves more fully, but not shifting on where we draw the lines,” is “a recipe for protracted and increasing acrimony with the [White House].” Id., at 573. “Given the bigger fish we have to fry with the Administration,” such as the EU-U. S. dispute over “data flows,” that did not “seem like a great place” for Facebook-White House relations “to be.” Ibid. So the platform was motivated to “explore some moves that we can make to show that we are trying to be responsive.” Ibid. That brainstorming resulted in the August 2021 rule changes. See supra, at 13, 19–20.

In sum, the officials wielded potent authority. Their communications with Facebook were virtual demands. And Facebook’s quavering responses to those demands show that it felt a strong need to yield.

For these reasons, I would hold that Hines is likely to prevail on her claim that the White House coerced Facebook into censoring her speech.

* �ă* �ă*

For months, high-ranking Government officials placed unrelenting pressure on Facebook to suppress Americans’ free speech. Because the Court unjustifiably refuses to address this serious threat to the First Amendment, I respectfully dissent.

Notes

[1]  Centers for Disease Control and Prevention, Deaths by Week and State, (last accessed June 21, 2024).

[2]  This includes information about the origin of the COVID–19 virus. When the pandemic began, Facebook began demoting posts supporting the theory that the virus leaked from a laboratory. See Interim Staff Report of the House Judiciary Committee, The Censorship-Industrial Complex: How Top Biden White House Officials Coerced Big Tech To Censor Americans, True Information, and Critics of the Biden Administration, p. 398 (May 1, 2024) (Committee Report), https://judiciary.house.gov/sites/evo-subsites / republicans-judiciary.house.gov/files/evo-media-document/Censorship-Industrial-Complex-WH-Report_Appendix.pdf. “In February 2021, in response to . . . tense conversations with the new Administration,” Facebook changed its policy to instead remove posts about the lab leak theory wholesale. Ibid.; accord, id., at 463 (Facebook executive explained that the platform removed these posts “[b]ecause we were under pressure from the administration and others to do more and it was part of the ‘more’ package”). But since then, both the Federal Bureau of Investigation and the Department of Energy have found that the theory is probably correct. See, e.g., A. Kaur & D. Diamond, FBI Director Says Covid–19 “Most Likely” Originated From Lab Incident, Washington Post (Feb. 28, 2023), ; J. Herb & N. Bertrand, US Energy Department Assesses Covid–19 Likely Resulted From Lab Leak, Furthering US Intel Divide Over Virus Origin, CNN (Feb. 27, 2023), . Facebook reversed its policy, and Mark Zuckerberg expressed regret that the platform had ever removed the posts: “This seems like a good reminder that when we compromise our standards due to pressure from an administration in either direction, we’ll often regret it later.” Committee Report 398.

[3]  See, e.g., J. Liedke & L. Wang, News Platform Fact Sheet, Pew Research Center (Nov. 15, 2023), ; A. Watson, Most Popular Platforms for Daily News Consumption in the United States as of August 2022, by Age Group, Statista (Jan. 4, 2024), .

[4]  C. Newton, Read the Full Transcript of Mark Zuckerberg’s Leaked Internal Facebook Meetings, The Verge (Oct. 1, 2019), .

[5]  For pending or potential legislation affecting internet platforms, see Congressional Research Service, C. Cho, L. Zhu, & K. Busch, Defining and Regulating Online Platforms (Aug. 25, 2023), .

[6]  E. Dwoskin, Massive Facebook Study on Users’ Doubt in Vaccines Finds a Small Group Appears To Play a Big Role in Pushing the Skepticism, Washington Post (Mar. 14, 2021), .

[7]  Notes recounting these calls were released by the House Judiciary Committee after the District Court entered the preliminary injunction and were published in a Committee Report. See Committee Report; Fed. Rule Evid. 201.

[8]  .

[9]  .

[10]  .

[11]  .

[12]  .

[13]  .

[14]  .

[15]  The Court discounts this evidence because Hines did not draw the same links in her briefing. See ante, at 20, n. 7. But we have an “independent obligation” to assess standing, Summers v. Earth Island Institute, 555 U.S. 488, 499 (2009), and a “virtually unflagging obligation” to exercise our jurisdiction if standing exists, Colorado River Water Conservation Dist. v. United States, 424 U.S. 800, 817 (1976). “[A] case like this one, where the record spans over 26,000 pages” and the plaintiffs have provided numerous facts, deserves some scrutiny before we simply brush standing aside. Ante, at 20, n. 7. As it happens, Hines has said enough to establish standing. First, she says that, at the behest of the White House, Facebook announced new measures to combat misinformation about COVID–19 and the vaccines. Second, she says that her Facebook pages fell under those policies. Third, she says that she suffered the penalties imposed by Facebook, such as demotion of her posts and pages. See 4 Record 1315; 78 id., at 25503. She may not explicitly say that the policy changes caused the penalties she experienced. But what theory makes more sense — that a user falling within Facebook’s amended policies was censored under those policies or that something else caused her injury?

[16]  See Meta, Policies, (last accessed June 19, 2024).

[17]  .

[18]  To obtain a preliminary injunction, Hines was required to establish that she is likely to succeed on the merits, that she would otherwise suffer irreparable harm, and that the equities cut in her favor. Winter v. Natural Resources Defense Council, Inc., 555 U.S. 7, 20 (2008). In a First Amendment case, the equities are bound up in the merits. See Elrod v. Burns, 427 U.S. 347, 373 (1976) (plurality opinion) (“The loss of First Amendment freedoms, for even minimal periods of time, unquestionably constitutes irreparable injury”). So I focus on Hines’s likelihood of success.

[19]  .

[20]  .

[21]  .

[22]  .

[23]  .

[24]  .

[25]  Webster’s International Dictionary of the English Language 191 (1902).

[26]  See D. Goodwin, The Bully Pulpit: Theodore Roosevelt, William Howard Taft, and the Golden Age of Journalism, pp. xi–xii (2013) (Goodwin).

[27]  .

[28]  .

[29]  .