Per Curiam.  In 2014, Michelle O’Connor-Ratcliff and T. J. Zane created public Facebook pages to promote their campaigns for election to the Poway Unified School District (PUSD) Board of Trustees. While O’Connor-Ratcliff and Zane (whom we will call the Trustees) both had personal Facebook pages that they shared with friends and family, they used their public pages for campaigning and issues related to PUSD. After they won election, the Trustees continued to use their public pages to post PUSD-related content, including board-meeting recaps, application solicitations for board positions, local budget plans and surveys, and public safety updates. They also used their pages to solicit feedback and communicate with constituents. Their Facebook pages described them as “Government Official[s]” and noted their official positions. O’Connor-Ratcliff also created a public Twitter page, which she used in much the same way.

 Christopher and Kimberly Garnier, who have children attending PUSD schools, often criticized the board of trustees. They began posting lengthy and repetitive comments on the Trustees’ social-media posts—for instance, nearly identical comments on 42 separate posts on O’Connor-Ratcliff ’s Facebook page and 226 identical replies within a 10-minute span to every tweet on her Twitter feed. The Trustees initially deleted the Garniers’ comments before blocking them from commenting altogether.

 The Garniers sued the Trustees under 42 U. S. C. §1983, seeking damages and declaratory and injunctive relief for the alleged violation of their First Amendment rights. At summary judgment, the District Court granted the Trustees qualified immunity as to the damages claims but allowed the case to proceed on the merits on the ground that the Trustees acted “under color of ” state law when they blocked the Garniers. §1983.

 The Ninth Circuit affirmed. It held that §1983’s state-action requirement was satisfied because there was a “close nexus between the Trustees’ use of their social media pages and their official positions.” 41 F. 4th 1158, 1170 (2022). The court cited its own state-action precedent, which holds that an off-duty state employee acts under color of law if she (1) “purports to or pretends to act under color of law”; (2) her “pretense of acting in the performance of [her] duties had the purpose and effect of influencing the behavior of others”; and (3) the “harm inflicted on plaintiff related in some meaningful way either to the officer’s governmental status or to the performance of [her] duties.” Ibid. (citing Naffe v. Frey, 789 F. 3d 1030, 1037 (CA9 2015); internal quotation marks and alterations omitted). Applying that framework, the court found state action based largely on the official “appearance and content” of the Trustees’ pages. 41 F. 4th, at 1171.

 We granted certiorari in this case and in Lindke v. Freed, ___ U. S. ___ (2024), to resolve a Circuit split about how to identify state action in the context of public officials using social media. 598 U. S. ___ (2023). Because the approach that the Ninth Circuit applied is different from the one we have elaborated in Lindke, we vacate the judgment below and remand the case to the Ninth Circuit for further proceedings consistent with our opinion in that case.

It is so ordered.

After pleading guilty to distributing at least 50 grams of methamphetamine, petitioner Mark Pulsifer faced a mandatory minimum sentence of 15 years in prison. At sentencing, he sought to take advantage of the “safety valve” provision of federal sentencing law, which allows a sentencing court to disregard the statutory minimum if a defendant meets five criteria. Among those is the requirement, set out in Paragraph (f)(1), that the sentencing court find that—

  (1) the defendant does not have—

  (A) more than 4 criminal history points, excluding any criminal history points resulting from a 1-point offense, as determined under the sentencing guidelines;

  (B) a prior 3-point offense, as determined under the sentencing guidelines; and

  (C) a prior 2-point violent offense, as determined under the sentencing guidelines.

 The Government argued that Pulsifer could not satisfy that requirement because he had two prior three-point offenses totaling six criminal-history points. In the Government’s view, each of those prior offenses disqualified him under Subparagraph B and the six total points disqualified him under Subparagraph A. But Pulsifer claimed he remained eligible. He pointed out that his criminal record lacked a two-point violent offense, as specified in Subparagraph C. And in his view, only the combination of the items listed in the subparagraphs could prevent him from getting safety-valve relief. The District Court agreed with the Government, and the Eighth Circuit affirmed.

Held: A defendant facing a mandatory minimum sentence is eligible for safety-valve relief under 18 U. S. C. §3553(f)(1) only if he satisfies each of the provision’s three conditions—or said more specifically, only if he does not have more than four criminal-history points, does not have a prior three-point offense, and does not have a prior two-point violent offense. Pp. 6–28.

 (a) Each party offers a grammatically permissible way to read Paragraph (f)(1). Under Pulsifer’s reading, the word “and” joins three features of a defendant’s criminal history into a single disqualifying characteristic; accordingly, a defendant is ineligible for the safety valve only if he has the items listed in Subparagraphs A, B, and C in combination. In the Government’s view, “and” connects three criminal-history conditions, all of which must be satisfied to gain safety-valve relief. In other words, the court must find the defendant does not have A, does not have B, and does not have C. Each of those readings is possible in the abstract. The choice between the two can sensibly be made only by examining the content of Paragraph(f)(1)’s three subparagraphs, including what they say, how they relate to each other, and how they fit with other pertinent law. Pp. 6–15.

 (b) The text and context of Paragraph (f)(1), as read against the Guidelines, yield just one plausible statutory construction. The paragraph creates an eligibility checklist, and specifies three necessary conditions for safety-valve relief. Reading the paragraph as Pulsifer does to set out a single condition—i.e., that the defendant not have the combination of the characteristics listed in Subparagraphs A, B, and C—would create two statutory difficulties that the Government’s reading does not. Pp. 15–23.

  (1) Pulsifer’s reading would render Subparagraph A superfluous because a defendant who has a three-point offense under Subparagraph B and a two-point offense under Subparagraph C will always have more than four criminal-history points under Subparagraph A. That reading leaves Subparagraph A with no work to do: removing it from the statute would make the exact same people eligible (and ineligible) for relief. That kind of superfluity, in and of itself, refutes Pulsifer’s reading. When a statutory construction “render[s] an entire subparagraph meaningless,” this Court has noted, the canon against surplusage applies with special force. National Assn. of Mfrs. v. Department of Defense, 583 U. S. 109, 128. That is particularly true when, as here, the subparagraph is so evidently designed to serve a concrete function. Pp. 15–20.

  (2) Pulsifer’s reading would also create a second problem related to Paragraph (f)(1)’s gatekeeping function. The Guidelines presume that defendants with worse criminal records—exhibiting recidivism, lengthy sentences, and violence—deserve greater punishment. Under the Government’s reading, Paragraph (f)(1) sorts defendants accord ingly. When the defendant has committed multiple non-minor offenses, he cannot get relief (Subparagraph A). And so too when he has committed even a single serious offense punished with a lengthy prison sentence (Subparagraph B) or one involving violence (Subparagraph C). Pulsifer’s reading, by contrast, would allow safety-valve relief to defendants with more serious records while barring relief to defendants with less serious ones. A defendant with a three-point offense and a two-point violent offense would be denied relief. But a defendant with multiple three-point violent offenses could get relief simply because he happens not to have a two-point violent offense.

 Contrary to Pulsifer’s view, that anomalous result cannot be ignored on the ground that a sentencing judge retains discretion to impose a lengthy sentence. If Congress thought it could always rely on sentencing discretion, it would not have created a criminal-history requirement in the first instance. Instead, it specified a requirement that allows such discretion to operate only if a defendant’s record does not reach a certain level of seriousness. Pulsifer’s construction of Paragraph (f)(1) makes a hash of that gatekeeping function. Pp. 20–23.

 (c) The uncontested fact that Congress amended Paragraph (f)(1) as part of the First Step Act to make safety-valve relief more widely available does not assist in interpreting the statutory text here. Both parties’ views of the paragraph widen the opportunity for safety-valve relief, and Pulsifer’s interpretation is not better just because it would allow more relief than the Government’s. “[N]o law pursues its . . . purpose[s] at all costs.” Luna Perez v. Sturgis Public Schools, 598 U. S. 142, 150. Here, where Congress did not eliminate but only curtailed mandatory minimums, the Court can do no better than examining Paragraph (f)(1)’s text in context to determine the exact contours of the defendants to whom Congress extended safety-valve relief. P. 26.

 (d) The Court rejects Pulsifer’s efforts to invoke the rule of lenity. Lenity applies only when a statute is genuinely ambiguous. For the reasons explained above, although there are two grammatically permissible readings of Paragraph (f)(1), in context its text is susceptible of only one possible construction. That leaves no role for lenity to play. Pp. 27–28.

39 F. 4th 1018, affirmed.

 Kagan, J., delivered the opinion of the Court, in which Roberts, C. J., and Thomas, Alito, Kavanaugh, and Barrett, JJ., joined. Gorsuch, J., filed a dissenting opinion, in which Sotomayor and Jackson, JJ., joined.

James Freed, like countless other Americans, created a private Facebook profile sometime before 2008. He eventually converted his profile to a public “page,” meaning that anyone could see and comment on his posts. In 2014, Freed updated his Facebook page to reflect that he was appointed city manager of Port Huron, Michigan, describing himself as “Daddy to Lucy, Husband to Jessie and City Manager, Chief Administrative Officer for the citizens of Port Huron, MI.” Freed continued to operate his Facebook page himself and continued to post prolifically (and primarily) about his personal life. Freed also posted information related to his job, such as highlighting communications from other city officials and soliciting feedback from the public on issues of concern. Freed often responded to comments on his posts, including those left by city residents with inquiries about community matters. He occasionally deleted comments that he considered “derogatory” or “stupid.”

  After the COVID–19 pandemic began, Freed posted about it. Some posts were personal, and some contained information related to his job. Facebook user Kevin Lindke commented on some of Freed’s posts, unequivocally expressing his displeasure with the city’s approach to the pandemic. Initially, Freed deleted Lindke’s comments; ultimately, he blocked him from commenting at all. Lindke sued Freed under 42 U. S. C. §1983, alleging that Freed had violated his First Amendment rights. As Lindke saw it, he had the right to comment on Freed’s Facebook page because it was a public forum. The District Court determined that because Freed managed his Facebook page in his private capacity, and because only state action can give rise to liability under §1983, Lindke’s claim failed. The Sixth Circuit affirmed.

Held: A public official who prevents someone from commenting on the official’s social-media page engages in state action under §1983 only if the official both (1) possessed actual authority to speak on the State’s behalf on a particular matter, and (2) purported to exercise that authority when speaking in the relevant social-media posts. Pp. 5–15.

 (a) Section 1983 provides a cause of action against “[e]very person who, under color of any statute, ordinance, regulation, custom, or usage, of any State” deprives someone of a federal constitutional or statutory right. (Emphasis added.) Section 1983’s “under color of” text makes clear that it is a provision designed as a protection against acts attributable to a State, not those of a private person. In the run-of-the-mill case, state action is easy to spot. Courts do not ordinarily pause to consider whether §1983 applies to the actions of police officers, public schools, or prison officials. Sometimes, however, the line between private conduct and state action is difficult to draw. In Griffin v. Maryland, 378 U. S. 130, for example, it was the source of the power, not the identity of the employer, which controlled in the case of a deputized sheriff who was held to have engaged in state action while employed by a privately owned amusement park. Since Griffin, most state-action precedents have grappled with whether a nominally private person engaged in state action, but this case requires analyzing whether a state official engaged in state action or functioned as a private citizen.

 Freed’s status as a state employee is not determinative. The distinction between private conduct and state action turns on substance, not labels: Private parties can act with the authority of the State, and state officials have private lives and their own constitutional rights—including the First Amendment right to speak about their jobs and exercise editorial control over speech and speakers on their personal platforms. Here, if Freed acted in his private capacity when he blocked Lindke and deleted his comments, he did not violate Lindke’s First Amendment rights—instead, he exercised his own. Pp. 5–8.

 (b) In the case of a public official using social media, a close look is definitely necessary to categorize conduct. In cases analogous to this one, precedent articulates principles to distinguish between personal and official communication in the social-media context. A public official’s social-media activity constitutes state action under §1983 only if the official (1) possessed actual authority to speak on the State’s behalf, and (2) purported to exercise that authority when he spoke on social media. The appearance and function of the social-media activity are relevant at the second step, but they cannot make up for a lack of state authority at the first. Pp. 8–15.

  (1) The test’s first prong is grounded in the bedrock requirement that “the conduct allegedly causing the deprivation of a federal right be fairly attributable to the State.” Lugar v. Edmondson Oil Co., 457 U. S. 922, 937 (emphasis added). Lindke’s focus on appearance skips over this critical step. Unless Freed was “possessed of state authority” to post city updates and register citizen concerns, Griffin, 378 U. S., at 135, his conduct is not attributable to the State. Importantly, Lindke must show more than that Freed had some authority to communicate with residents on behalf of Port Huron. The alleged censorship must be connected to speech on a matter within Freed’s bailiwick. There must be a tie between the official’s authority and “the gravamen of the plaintiff’s complaint.” Blum v. Yaretsky, 457 U. S. 991, 1003.

 To misuse power, one must possess it in the first place, and §1983 lists the potential sources: “statute, ordinance, regulation, custom, or usage.” Determining the scope of an official’s power requires careful attention to the relevant source of that power and what authority it reasonably encompasses. The threshold inquiry to establish state action is not whether making official announcements could fit within a job description but whether making such announcements is actually part of the job that the State entrusted the official to do. Pp. 9–12.

  (2) For social-media activity to constitute state action, an official must not only have state authority, he must also purport to use it. If the official does not speak in furtherance of his official responsibilities, he speaks with his own voice. Here, if Freed’s account had carried a label—e.g., “this is the personal page of James R. Freed”—he would be entitled to a heavy presumption that all of his posts were personal, but Freed’s page was not designated either “personal” or “official.” The ambiguity surrounding Freed’s page requires a fact-specific undertaking in which posts’ content and function are the most important considerations. A post that expressly invokes state authority to make an announcement not available elsewhere is official, while a post that merely repeats or shares otherwise available information is more likely personal. Lest any official lose the right to speak about public affairs in his personal capacity, the plaintiff must show that the official purports to exercise state authority in specific posts. The nature of the social-media technology matters to this analysis. For example, because Facebook’s blocking tool operates on a page-wide basis, a court would have to consider whether Freed had engaged in state action with respect to any post on which Lindke wished to comment. Pp. 12–15.

37 F. 4th 1199, vacated and remanded.

 Barrett, J., delivered the opinion for a unanimous Court.