COMPUTER & COMMUNICATIONS INDUSTRY ASSOCIATION

v.

ATTORNEY GENERAL, STATE OF FLORIDA

Appeal from the United States District Court for the Northern District of Florida D.C. Docket No. 4:24-cv-00438-MW-MAF ____________________ USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 2 of 59

Before NEWSOM, LUCK, and TJOFLAT, Circuit Judges. NEWSOM, Circuit Judge: For better or worse, the proliferation of social media has transformed American life. Unsurprisingly, it has also generated “a raft of public policy issues.” Moody v. NetChoice, LLC, 603 U.S. 707, 716 (2024). One of the most pressing is how and in what ways social-media use affects minors. According to the State of Florida, social-media companies seed their platforms with “addictive features”—push notifications, auto-play, infinite scrolling, etc.—in order to make them, well, addictive. These features, the state says, are “psychological hooks” that “overstimulate the reward center in the brain” and “foster compulsive overuse.” Br. of Appellant at 4 (citations omitted). Florida insists that children are particularly susceptible to these “addictive features” because their brains are still developing. In particular, the state contends, the features contribute to “[c]ompulsive or uncontrollable” social-media use among children and threaten a “youth mental health crisis.” Id. at 1, 5. Children face “‘sleep problems, attention problems, and feelings of exclusion,’ all of which put [them] at risk of ‘depression, anxiety, and neuroticism.’” Id. at 5 (citations omitted). In an effort to protect children from social-media addiction and its ensuing effects, Florida passed HB³, which imposes age-restriction requirements on platforms that use “addictive features.” Fla. Stat. § 501.1736(1)(e). In particular, HB³ requires platforms USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 3 of 59

within its ambit to prohibit anyone under the age of 14 from creating an account, and to prohibit anyone else under 16 from doing so without parental consent. Id. § 501.1736(2)–(3). The Computer & Communications Industry Association and NetChoice, both Internet trade organizations—to which, collectively, we will hereinafter refer as “CCIA”—brought facial and as-applied First Amendment challenges to HB3. The district court concluded that HB³ was likely unconstitutional and preliminarily enjoined Florida from enforcing it. We stayed the injunction pending appeal. Before us, Florida contends (1) that CCIA lacks standing to assert the First Amendment rights of minors, (2) that the district court should have abstained from deciding the case under Younger v. Harris, 401 U.S. 37 (1971), and (3) that HB³ satisfies First Amendment scrutiny.

Our analysis proceeds in three parts. First, as threshold matters, we hold that CCIA has standing to challenge HB³ and that Younger abstention doesn’t foreclose its claim. Second, we address the merits of CCIA’s as-applied challenge, and we hold that HB³ likely passes First Amendment muster because it is a content-neutral speech regulation that satisfies intermediate scrutiny. Finally, we reject CCIA’s facial challenge to HB³ under the principles articulated in Moody.¹

other states restricting minors’ use of social-media platforms. See, e.g., NetChoice v. Carr, 789 F. Supp. 3d 1200 (N.D. Ga. 2025) (Georgia), appeal docketed, No. 25-12436 (11th Cir. July 16, 2025); NetChoice, LLC v. Skrmetti, No. 25- USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 4 of 59

I

A

We begin with a primer on HB3. The law defines a “social media platform” as “an online forum, website, or application” (1) that “[a]llows users to upload content or view the content or activity of other users”; (2) on which at least 10% of daily active users under 16 years old spend an average of at least two hours per day; (3) that “[e]mploys algorithms that analyze user data or information on users to select content for” them; and (4) that “[h]as any of the following addictive features”: “[i]nfinite scrolling,” “[p]ush notifications,” displays of “personal interactive metrics,” “[a]utoplay,” or “[l]ive-streaming.” Fla. Stat. § 501.1736(1)(e). HB³ defines each “addictive feature” as follows: • “Infinite scrolling” refers to either (1) the “[c]ontinuous[] loading” of content without the need to open a separate page, or (2) the “[s]eamless” loading of content “with no visible or apparent end or page breaks.”

5660, 2026 WL 2546423, at *2 (6th Cir. Aug. 28, 2026) (Tennessee); Comput. & Commc’ns Indus. Ass’n v. Paxton, 184 F. 4th 420, 437 (5th Cir. 2026) (Texas); NetChoice, LLC v. Yost, 180 F. 4th 268 (6th Cir. 2026) (Ohio); NetChoice, LLC v. Bonta, 152 F. 4th 1002 (9th Cir. 2025) (California); NetChoice, L.L.C. v. Fitch, 134 F. 4th 799 (5th Cir. 2025) (Mississippi); NetChoice v. Murrill, 812 F. Supp. 3d 594 (M.D. La. 2025) (Louisiana); NetChoice, LLC v. Griffin, No. 5:23-CV-5105, 2025 WL 978607 (W.D. Ark. Mar. 31, 2025) (Arkansas); NetChoice, LLC v. Reyes, 748 F. Supp. 3d 1105 (D. Utah 2024) (Utah).

USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 5 of 59

• “Push notifications” are “alerts” sent by the platform “to inform a user about specific activities or events related to the user’s account.” • “Personal interactive metrics” refer to “[d]isplays . . . indicat[ing] the number of times other users have clicked a button to indicate their reaction to content or have shared or reposted the content”—think “likes,” “comments,” “reposts,” “shares,” etc. • “Auto-play” is a feature that enables a video to “play without the user first clicking on the video or on a play button for that video.” • “Live-streaming” is a “function that allows a user or advertiser to broadcast live video content in real-time.” Id. § 501.1736(1)(e)(4)(a)–(e). HB³ exempts from its definition of “social media platform” any “online service, website, or application where the exclusive function is e-mail or direct messaging consisting of text, photographs, pictures, images, or videos shared only between the sender and the recipients.” Id. HB³ requires covered platforms (1) to prohibit minors under the age of 14 from creating accounts and (2) to prohibit 14- and 15year-olds from creating accounts without parental consent. Id. § 501.1736(2)(a), (3)(a). If an account holder is under 14, or is 14 to 15 but doesn’t have parental consent, the platform must terminate her account. Id. § 501.1736(2)(b)(1), (3)(b). USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 6 of 59

B

On October 28, 2024, shortly before HB³ took effect, CCIA brought facial and as-applied First Amendment challenges to HB³’s pertinent provisions. The next day, CCIA moved for a preliminary injunction to prevent HB³’s enforcement. A flurry of docket activity ensued. (The procedural recitation that follows may seem pretty tedious, but it’s relevant to the Younger issue before us.) Lawyers filed notices of appearance and withdrawal, as well as pro hac vice motions, on which the district court ruled. Florida filed a motion to dismiss, to which CCIA responded. The state also filed an opposition to CCIA’s preliminary-injunction motion, and CCIA filed a reply in support. In addition, Florida deposed four industry witnesses. See Cleland Dep., Dkt. No. 51-4 (NetChoice’s General Counsel); Schruers Dep., Dkt. No. 51-5 (CCIA’s President and CEO); Boyle Dep., Dkt. No. 51-6 (Snap’s Senior Director of Product Management); Veitch Dep., Dkt. No. 51-7 (YouTube’s Director of Government Affairs and Public Policy across the Americas). Following a hearing, the district court denied CCIA’s preliminary-injunction motion on associational standing grounds. The court held that CCIA had failed to show that any of its members’ asserted injuries were “actual or imminent” and “fairly traceable” to the state because it didn’t allege “that at least one specific member of [CCIA] [was] likely regulated by the law.” Dist. Ct. Order, Mar. 13, 2025, at 6. The court also granted Florida’s motion to dismiss CCIA’s complaint. CCIA promptly filed an amended complaint that primarily sought to remedy the standing defects that the district court had identified and renewed its preliminary-injunction USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 7 of 59

motion. The court held a scheduling conference and set a briefing schedule. In the meantime, CCIA deposed three of the state’s expert witnesses. See Twenge Dep. 32–34, 53–57, Dkt. No. 63-1 (opining that using social media causes, or at least correlates with, harmful outcomes for children); Allen Dep. 12–16, 58–139, Dkt. No. 63- 2 (explaining how age-verification systems function); Alter Dep. 27–54, Dkt. No. 63-3 (describing a theory of how design features of social-media platforms trigger behavioral addiction). On April 21, 2025, before the district court had ruled on CCIA’s renewed preliminary-injunction motion, Florida filed an enforcement action in state court against Snap, a social-media company and member of CCIA’s co-plaintiff NetChoice, alleging violations of HB3. Snap removed the enforcement action to federal district court.² The district court then granted CCIA’s renewed preliminary-injunction motion. As relevant here, the court determined (1) that CCIA had standing, (2) that Younger didn’t require dismissal of CCIA’s suit, and (3) that although HB³ was likely a content-neutral speech regulation, it was likely insufficiently tailored to survive

Inc., No. 3:25-cv-676 (N.D. Fla. May 21, 2025), Dkt. No. 1. On the same day that we heard oral argument in this case, we heard argument in the appeal from the district court’s order declining to remand the enforcement action to state court. See Off. of the Att’y Gen., State of Fla., Dep’t of Legal Affs. v. Snap Inc., No. 25-12814.

USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 8 of 59

intermediate scrutiny. The court thus enjoined Florida from enforcing the challenged provisions of HB3.

We stayed the preliminary injunction pending appeal. In our stay order, we agreed with the district court (albeit tentatively) that CCIA had standing, that Younger abstention didn’t apply, and that the statute was likely content neutral. Nevertheless, we concluded (again, tentatively) that HB³ was sufficiently tailored to satisfy intermediate scrutiny. We therefore granted the stay on the ground that Florida was likely to succeed on the merits. Following oral argument of the appeal of the district court’s preliminary-injunction order, we directed the parties to submit supplemental briefing on the questions (1) whether CCIA has standing and (2) whether the district court decided CCIA’s as-applied challenge.

II

Before diving into the merits, we need to address two threshold questions: (1) whether CCIA has standing to pursue its First Amendment challenge, and (2) whether the district court should have abstained from deciding this case under Younger. For reasons we’ll explain, we answer those questions yes and no, respectively.

USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 9 of 59

A

First, standing.³ “Article III standing is a ‘bedrock constitutional requirement.’” Food & Drug Admin. v. All. for Hippocratic Med., 602 U.S. 367, 378 (2024) (quoting United States v. Texas, 599 U.S. 670, 675 (2023)). To establish standing, “a plaintiff must demonstrate (i) that she has suffered or likely will suffer an injury in fact, (ii) that the injury likely was caused or will be caused by the defendant, and (iii) that the injury likely would be redressed by the requested judicial relief.” Id. at 380.

CCIA asserts standing to vindicate the First Amendment rights of its members—i.e., the social-media platforms. The Supreme Court has recognized that “an organization may sue to redress its members’ injuries” under what is known as “associational standing.” United Food & Com. Workers Union Loc. 751 v. Brown Grp., 517 U.S. 544, 552 (1996). It’s a “form of third-party standing” that arguably “relax[es] both the injury and redressability requirements for Article III standing.” All. for Hippocratic Med., 602 U.S. at 398– 99 (Thomas, J., concurring).

Under associational-standing principles, an organization may sue on its members’ behalf when (1) “its members would otherwise have standing to sue in their own right,” (2) “the interests it seeks to protect are germane to the organization’s purpose,” and (3) “neither the claim asserted nor the relief requested requires the

participation of individual members in the lawsuit.” Brown Grp., 517 U.S. at 553 (quoting Hunt v. Wash. State Apple Advert. Comm’n, 432 U.S. 333, 343 (1977)).¹ To satisfy the associational-standing test’s first requirement, at least one of CCIA’s members must itself meet the three Article III standing elements: (1) injury in fact, (2) causation, and (3) redressability. See id.

An “injury in fact” is “an invasion of a legally protected interest which is (a) concrete and particularized, and (b) actual or imminent, not conjectural or hypothetical.” Lujan v. Defs. of Wildlife, 504 U.S. 555, 560 (1992) (citation modified). In the pre-enforcement-challenge context, a “plaintiff satisfies the injury-in-fact requirement where he alleges an intention to engage in a course of conduct arguably affected with a constitutional interest, but proscribed by a statute, and there exists a credible threat of prosecution thereunder.” Susan B. Anthony List v. Driehaus, 573 U.S. 149, 159 (2014) (citation modified). In this circuit, “[t]he injury requirement is most loosely applied—particularly in terms of how directly the injury must result from the challenged government action—where [F]irst [A]mendment rights are involved, because of the fear that free speech will be chilled even before the law, regulation, or policy is enforced.” Speech First, Inc. v. Cartwright, 32 F. 4th 1110, 1120 (11th Cir. 2022). A First Amendment plaintiff has standing if “the operation or enforcement of the government policy would cause a USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 11 of 59

reasonable would-be speaker to self-censor—even where the policy falls short of a direct prohibition against the exercise of First Amendment rights.” Id. (citation modified). “The fundamental question under our precedent . . . is whether the challenged policy ‘objectively chills’ protected expression.” Id. At least a handful of CCIA’s members satisfy the injury-infact requirement. The alleged conduct in which Snap, YouTube, and Meta engage—and intend to continue to engage—is “arguably affected with a constitutional interest” that is “proscribed by” HB3. Susan B. Anthony List, 573 U.S. at 159. CCIA contends that its member platforms are likely to suffer First Amendment injuries because HB³ will hinder their ability to communicate with their users through “curated compilations of third-party speech,” Supp. Br. of Appellees at 3–4, and features such as “[p]ush notifications” and “personal interactive metrics,” Am. Compl. ¶¶ 23, 36, 49. “[C]ompiling and curating others’ speech” is an “expressive activity” protected by the First Amendment. Moody, 603 U.S. at 731. The Moody Court concluded, based on the record in that case, that Facebook’s and YouTube’s decisions about “which third-party content those feeds will display, or how the display will be ordered and organized,” are “expressive choices” meriting First Amendment protection. Id. at 740. The same reasoning applies here. CCIA has alleged that Snap, YouTube, and Meta (through Facebook and Instagram) curate content “that they think will be particularly relevant to” their users and choose that content for “a particular user based in part on the content that the user has interacted with in the past.” Am. Compl. ¶ 48; see also id. ¶¶ 22, 35. Therefore, by restricting USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 12 of 59

some users’ ability to create accounts, HB³ would limit the platforms’ ability to engage in the expressive activity of delivering curated content to those users.

Moreover, at least some of CCIA’s members face a “credible threat of prosecution.” Susan B. Anthony List, 573 U.S. at 159 (citation omitted). We’ve said that “when a plaintiff challenges a law soon after it was enacted and the state vigorously defends the law in court, we can infer a credible threat of prosecution exists.” Shen v. Comm’r, Fla. Dep’t of Agric. & Consumer Servs., 158 F. 4th 1227, 1242 (11th Cir. 2025) (citation modified). That’s what happened here: The state has been nothing if not vigorous. And the record shows that Snap and YouTube, for example, meet HB³’s coverage criteria. The enforcement of HB³ would not only objectively chill these platforms from delivering curated content to the younger users HB³ covers, see Speech First, Inc., 32 F. 4th at 1120, but would also pressure them into instituting “significant and costly compliance measures” to implement age-verification and parental-verification systems. Am. Compl. ¶ 21 (Snap); id. ¶ 34 (YouTube); id. ¶ 47 (Meta). Cf. Virginia v. Am. Booksellers Ass’n, 484 U.S. 383, 392 (1988) (holding that the plaintiffs met the injury-in-fact requirement because, “if their interpretation of the statute is correct, [they] will have to take significant and costly compliance measures or risk criminal prosecution”). CCIA has thus shown that at least one of its members is likely to suffer an injury in fact. The remaining requirements for Article III standing—causation and redressability—are clearly satisfied by at least one CCIA USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 13 of 59

member. The “defendant’s conduct”—here, Florida’s enactment of HB³—is the cause of the platform members’ First Amendment injuries, and relief in the form of an injunction prohibiting HB³’s enforcement would doubtlessly redress them. Accordingly, CCIA has met the first prong of the associational-standing test by showing that at least one of its members has Article III standing.² As for the second prong of the associational-standing test, we agree with the district court—and Florida doesn’t contest—that the interests that CCIA “aims to protect are germane to [its] purposes.” Brown Grp., 517 U.S. at 553. 3 CCIA also satisfies the third and final prong—namely, that its claim can be resolved and requested relief can be granted without the participation of its individual members. Brown Grp., 517 U.S. at 553. The Supreme Court has called the individual-participation element a “judicially self-imposed limit” that aims to serve “administrative convenience and efficiency.” Id. at 555–57. Unlike the first two prongs, which are “constitutional necessit[ies],” this third prong is “prudential.” Id. at 555–56. Meeting it rebuts the “background presumption” that a litigant (here, CCIA) may not assert the rights of absent third parties (here, its member platforms). Id. at 557.

USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 14 of 59

In similar circumstances, a number of other courts have held that CCIA’s co-plaintiff, NetChoice, satisfies the associationalstanding test’s individual-participation prong. NetChoice L.L.C. v. Fitch, 134 F. 4th 799, 805 (5th Cir. 2025) (“NetChoice’s ‘claims can be proven by evidence from representative injured members’ and ‘the participation of [certain] individual members does not thwart associational standing.’”); CCIA v. Paxton, 747 F. Supp. 3d 1011, 1030 (W.D. Tex. 2024), aff’d in relevant part, 184 F. 4th 420, 445 (5th Cir. 2026); NetChoice v. Murrill, 812 F. Supp. 3d 594, 629–31 (M.D. La. 2025); NetChoice, LLC v. Reyes, 748 F. Supp. 3d 1105, 1118–19 (D. Utah 2024); NetChoice, LLC v. Griffin, No. 5:23-CV-5105, 2023 WL 5660155, at *10 (W.D. Ark. Aug. 31, 2023). We agree with them. The state argues that CCIA doesn’t have associational standing because, on the merits, its First Amendment claim requires “numerous factual determinations about individual platforms.” Supp Br. of Appellant at 6. For support, it cites Harris v. McRae, 448 U.S. 297, 321 (1980), and Georgia Cemetery Ass’n v. Cox, 353 F. 3d 1319, 1322 (11th Cir. 2003). But both cases are distinguishable. In Harris, a pro-choice division of a religious organization asserted that federal abortion-funding restrictions violated the First Amendment’s Free Exercise Clause. 448 U.S. at 321. The Supreme Court held that the group couldn’t proceed under an associational-standing theory because adjudicating the free-exercise issues would necessarily entail an inquiry into whether and to what extent the restrictions operated coercively on particular members. Id. In Cox, a cemetery association brought a Takings Clause challenge to a law capping fees for the transfer of burial rights. 353 F. 3d at 1322. We USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 15 of 59

held that the group couldn’t assert associational standing because resolving its takings claim required the participation of its members. “[T]he economic impact,” we reasoned, “will vary depending upon the economic circumstances of” particular cemeteries. Id. Generally, whether a violation occurred in the regulatory-takings context requires an individualized inquiry into whether each affected party was denied the compensation it was due. Tahoe-Sierra Pres. Council v. Tahoe Reg’l Plan. Agency, 535 U.S. 302, 322 (2002) (characterizing regulatory-takings jurisprudence as entailing “‘essentially ad hoc, factual inquiries,’ designed to allow ‘careful examination and weighing of all the relevant circumstances’” (citation omitted)).

This case is different. While free-exercise and takings claims require consideration of the law’s effects on individual parties, CCIA’s free-speech claim, which it brings on behalf of its member platforms, turns on whether HB³ implicates the First Amendment, whether it is content based or content neutral, and whether it is appropriately tailored to the asserted state interest—inquiries that don’t depend on fact-intensive determinations regarding the member platforms. Nor does CCIA’s requested remedy require the individual members’ participation—for the simple reason that it seeks declaratory and injunctive relief, rather than damages. See Brown Grp., 517 U.S. at 553–54 (“If in a proper case the association seeks a declaration, injunction or some other form of prospective relief, it can reasonably be supposed that the remedy, if granted, will inure to the benefit of those members of the association actually injured.” (citation modified)).

USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 16 of 59

The state separately argues that First Amendment “claim[s] related to algorithmic speech” are “fact intensive” and therefore will “surely vary from platform to platform.” Supp. Br. of Appellant at 10 (quoting NetChoice, LLC v. Bonta, 152 F. 4th 1002, 1014 (9th Cir. 2025)). Not necessarily. To be sure, in Bonta, on which the state relies, the Ninth Circuit held that a challenge brought by NetChoice required the participation of individual member platforms. But the law and the challenge there don’t resemble those here. In Bonta, NetChoice sued to enjoin the enforcement of a law that restricted minors’ access to “personalized feeds” generated by the platforms’ content-recommendation algorithms. 152 F. 4th at 1010, 1014. Based on NetChoice’s acknowledgment that “each of its members” and their algorithms were “unique,” the Ninth Circuit concluded that the law would implicate the First Amendment only with respect to platforms whose algorithms were “expressive.” Id. at 1014. The determination whether each platform’s algorithm was expressive, the court concluded, was “fact intensive” and required individual members’ participation. Id. HB³ is very different; its restrictions are tethered not to each platform’s unique algorithmic design but to so-called addictive features that are present across platforms.

We are satisfied that CCIA has satisfied the third and final prong of associational standing. * * * USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 17 of 59

For these reasons, we hold that CCIA has associational standing and may bring this action on behalf of its members.⁴

B

We next address whether Younger v. Harris, 401 U.S. 37 (1971), required the district court to abstain from ruling on CCIA’s

tended that it also has third-party standing to vindicate the First Amendment rights of its members’ users. See, e.g., Am. Compl. ¶ 50 (“CCIA and NetChoice have standing to assert . . . the First Amendment rights of their members’ current and prospective users.”); id. ¶¶ 7, 8 (same); Br. of Appellees at 43 (arguing that CCIA and NetChoice have “‘prudential standing’ to assert the First Amendment rights of their members’ users”). In its supplemental brief, though, CCIA seems to have retreated from its original, standing-based position, and now it says instead that its contentions regarding users’ rights were meant only to supplement the merits inquiry. See Supp. Br. of Appellees at 15 (calling the assertion of prudential standing a “misunderstanding”); id. (“Plaintiffs invoke users’ rights not to get into court or expand their universe of claims, but because users are a necessary component of the requisite First Amendment inquiry[.]”). In any event, CCIA’s assertion of third-party standing to pursue its members’ users’ rights runs more to its facial overbreadth claim than its as-applied claim. But for reasons we explain in Part IV, we conclude that, at least on the record as it currently exists, CCIA hasn’t made the showing necessary to proceed with any facial challenge. If CCIA wishes to continue to assert the rights of its members’ users in connection with a facial overbreadth challenge, the district court may address those issues then. One more standing-related loose end: CCIA separately asserts thirdparty standing to pursue its members’ users’ rights on jus tertii grounds, which could apply to its as-applied claim. We needn’t address this argument because we find that CCIA has associational standing. USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 18 of 59

preliminary-injunction motion in light of Florida’s state-court enforcement action against Snap, a member of NetChoice.⁵ “[A]bstaining from the ‘exercise of federal jurisdiction is the exception, not the rule.’” New Ga. Project, Inc. v. Att’y Gen. of Ga., 106 F. 4th 1237, 1241 (11th Cir. 2024) (quoting Colo. River Water Conservation Dist. v. United States, 424 U.S. 800, 813 (1976)). Accordingly, only the “clearest of justifications merits abstention.” Id. (quoting Jackson-Platts v. Gen. Elec. Cap. Corp., 727 F. 3d 1127, 1140 (11th Cir. 2013)). In Younger, the Supreme Court recognized a “limited exception to the usual rule.” Id.

Under Younger, a federal court must abstain if (1) “there is an ongoing state judicial proceeding,” (2) “the state proceeding implicates important state interests,” and (3) “there is an adequate opportunity in the state proceeding to raise constitutional challenges.” Id. at 1242 (citation modified) (quoting Middlesex Cnty. Ethics Comm. v. Garden State Bar Ass’n, 457 U.S. 423, 432 (1982)). The question before us involves the first requirement—whether Florida’s enforcement action against Snap was “ongoing” in the relevant sense. Id.

Perhaps counterintuitively, a state proceeding is deemed to be “ongoing” for Younger purposes “even if it’s not actually pending when the federal suit is filed—provided that it’s instituted . . . ‘before any proceedings of substance on the merits have taken place

v. Att’y Gen. of Fla., 377 F. 3d 1258, 1262 (11th Cir. 2004). USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 19 of 59

in federal court.’” Id. at 1243 (quoting Hicks v. Miranda, 422 U.S. 332, 349 (1975)) (emphasis in original). That means that if the federal suit is still in an “embryonic stage” and “no contested matter has been decided” by the federal court by the time the state brings its enforcement action, the latter will be treated as if it were “ongoing”—meaning that the federal court must abstain from further litigation. Tokyo Gwinnett v. Gwinnett Cnty., 940 F. 3d 1254, 1268 (11th Cir. 2019).

Whether “proceedings of substance on the merits” have occurred in a federal suit entails a “fact-specific assessment of the circumstances.” Id. at 1272. That assessment depends on “the time that the district court has spent considering the case, any motions ruled on, any discovery, the number of conferences [or hearings] held, and any change in the parties’ position as a result of the federal litigation.” Id. “[T]he filing of motions” is also relevant, “even if the district court did not rule” on them. Id. And even “significant and substantial procedural steps” can qualify as “proceedings of substance on the merits” sufficient to preclude abstention. Id. at 1271–72.

Florida brought its enforcement action against Snap nearly six months after CCIA initially filed its lawsuit, and during that span, “proceedings of substance on the merits” undoubtedly occurred in federal court. Let’s recap: (1) Florida moved to dismiss CCIA’s action, and (2) CCIA responded; (3) Florida filed an opposition to CCIA’s preliminary-injunction motion, and (4) CCIA re- USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 20 of 59

plied; (5) the district court held a hearing on CCIA’s motion, a substantial portion of which was dedicated to a discussion of the merits of CCIA’s claim, see Prelim. Inj. Hr’g Tr. 21–41, 68–92, 101–114, 123–135, Dkt. No. 71; (6) the court then denied CCIA’s motion (albeit on standing grounds); (7) CCIA filed an amended complaint and (8) renewed its preliminary-injunction motion; and (9) the district court convened a status conference and (10) issued a briefing schedule.

Along the way, (11) the parties conducted limited discovery, including several depositions, all of which bore on the merits to some extent. In particular, the state deposed four industry witnesses, see Cleland Dep., Dkt. No. 51-4 (testifying about NetChoice’s members, their business practices, and their age-verification systems); Schruers Dep., Dkt. No. 51-5 (testifying about CCIA’s members, their business practices, and their parental controls); Boyle Dep., Dkt. No. 51-6 (testifying about Snapchat’s features and parental controls); Veitch Dep., Dkt. No. 51-7 (testifying about YouTube’s features and age-based policies and systems). So too, CCIA deposed three of Florida’s expert witnesses, see Twenge Dep. 32–34, 53–57, Dkt. No. 63-1 (testifying about how social-media features trigger harmful behavior in children); Allen Dep. 12– 16, 58–139, Dkt. No. 63-2 (testifying about how age-verification systems function); Alter Dep. 27–54, Dkt. No. 63-3 (testifying about research on behavioral addiction and features of digital platforms that make them difficult to resist).

USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 21 of 59

To recount this history, we think, is to answer the Younger question: By the time Florida brought its enforcement action against Snap, CCIA’s federal case had passed the “embryonic stage.” Tokyo Gwinnett, 940 F. 3d at 1268. The district court did not abuse its discretion in declining to abstain under Younger.

III

On, then, to the merits.⁶ As already explained, CCIA presented, and the district court addressed, both facial and as-applied challenges to HB3.7 We will consider CCIA’s as-applied challenge

discretion” and the underlying questions of law de novo. Noble Prestige Ltd. v. Galle, 83 F. 4th 1366, 1374 (11th Cir. 2023).

court’s preliminary injunction might be understood as invalidating HB³ only on its face. See Dist. Ct. Order, June 3, 2025, at 57 n.33 (“[A]lthough Plaintiffs’ amended complaint and motion for preliminary injunction do not expressly state that they are raising a facial challenge, their briefing indicates that they are . . . . Regardless, even if Plaintiffs had not brought a facial challenge, this Court would be permitted to evaluate the law’s constitutionality on its face.”); id. at 58 (“Defendant Uthmeier must take no steps to enforce [the challenged provisions of HB³] until otherwise ordered.”). Notably, though, in their supplemental briefs before us, the parties agree that the district court addressed CCIA’s as-applied claim, as well. See Supp. Br. of Appellant at 20–21; Supp. Br. of Appellees at 19–20. We agree with them. Understood in the context of the district court’s entire order—and even more so in the light of the parties’ competing contentions before that court—we think it clear that the court preliminarily enjoined HB³ both on its face and as applied to CCIA’s members. See, e.g., Dist. Ct. Order, June 3, 2025, at 25 (“This Court begins by noting that Plaintiffs bring both facial and as-applied challenges to Florida’s law . . . .”); id. at 28 (observing that much of the USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 22 of 59

first, and then deal with its facial challenge—which presents unique issues—in Part IV. Our evaluation of CCIA’s as-applied challenge proceeds in three parts. First, we ask whether HB³ implicates the First Amendment at all. Second, if it does, we must determine whether HB³ is content based or content neutral. And finally, we must decide whether HB³ survives the requisite level of scrutiny.

A

A law implicates the First Amendment—and thus triggers constitutional scrutiny of some sort—if it either (1) “directly regulate[s] protected expressive activity” or (2) regulates even non-expressive conduct in a way that “impose[s] a disproportionate burden upon those engaged in protected First Amendment activities.” TikTok, Inc. v. Garland, 604 U.S. 56, 68 (2025) (per curiam) (quoting Arcara v. Cloud Books, Inc., 478 U.S. 697, 703–04 (1986)). Florida contends that HB³ “regulates product design, not children’s speech,” and, accordingly, doesn’t even implicate the First Amendment. Br. of Appellant at 27. In particular, the state says, HB³ restricts only the creation of accounts, not social-media

First Amendment analysis “is the same for both the facial and as-applied challenges,” and explaining the district court’s view that only at the tailoring stage does a court “separately consider[] whether Plaintiffs have carried their burden to sustain a facial challenge”); id. at 52–53 (“[T]his Court concludes that Plaintiffs are substantially likely to succeed not just in their as-applied challenge but in their facial challenge as well.”).

USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 23 of 59

access writ large, and thus regulates only “commercial transaction[s]”—namely, a platform’s ability to “enter[] into a contract” with children of a certain age. Id. at 28.

Even accepting the state’s characterization, it seems clear enough to us that HB³ “disproportionate[ly] burden[s]” CCIA’s member platforms’ “protected First Amendment activities,” Tik- Tok, 604 U.S. at 68—namely, “compiling and curating others’ speech” for their users, Moody, 603 U.S. at 731. In particular, HB³ burdens platforms’ ability to communicate with their users by requiring them (1) to prohibit minors under 14 years of age from creating or maintaining accounts and (2) to secure parental consent before communicating with 14- and 15-year-olds. We have little trouble concluding that HB³’s restrictions implicate the First Amendment—and thus trigger constitutional scrutiny. The trickier question, to which we turn next: What kind of scrutiny?

B

First Amendment doctrine distinguishes between speech restrictions that are “content based” and those that are “content neutral.” Free Speech Coal. v. Paxton, 606 U.S. 461, 471 (2025). A content-based regulation triggers “strict” constitutional scrutiny, meaning that it must represent “the least restrictive means” of accomplishing a compelling governmental interest. Id. (quoting McCullen v. Coakley, 573 U.S. 464, 478 (2014)). A content-neutral regulation, by contrast, triggers only “intermediate” scrutiny, USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 24 of 59

meaning (although the formulations vary) that it need only “advance[] important governmental interests unrelated to the suppression of free speech and . . . not burden substantially more speech than necessary to further those interests.” TikTok, 604 U.S. at 70 (quoting Turner Broad. Sys., Inc. v. FCC, 520 U.S. 180, 189 (1997) (Turner II)). For reasons we will explain, we hold that HB³ is a content-neutral speech regulation. Accordingly, the law is subject to intermediate, rather than strict, scrutiny.¹ A law is content based if it “target[s] speech based on its communicative content.” TikTok, 604 U.S. at 70 (quoting Reed v. Town of Gilbert, 576 U.S. 155, 163 (2015)). To be clear, a law can be content based either on its face or in its justification. A “law is content based on its face if it ‘applies to particular speech because of the topic discussed or the idea or message expressed.’” Id. (quoting Reed, 576 U.S. at 163); see also Free Speech Coal., 606 U.S. at 504 (Kagan, J., dissenting) (observing that a law is content based if it “draws distinctions based on the topic, subject matter, idea, or message expressed” (citation modified)). Alternatively, even a facially content-neutral law can be content-based if “it cannot be justified without reference to the content of the regulated speech or was adopted by the government because of disagreement with the message the speech conveys.” TikTok, 604 U.S. at 70–71 (quoting Reed, 576 U.S. at 164 (citation modified)).

And the flip side? A law is facially content neutral if it doesn’t “target particular speech based upon its content”—or, put another USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 25 of 59

way, doesn’t “impose a ‘restriction, penalty, or burden’ by reason of content.” Id. at 71 (quoting Turner Broad. Sys., Inc. v. FCC, 512 U.S. 622, 644 (1994) (Turner I)). A good proxy for content neutrality: Can the speaker “‘avoid or mitigate’ the effects of the [law] by altering” the substance of his or her expression? Id. (quoting Turner I, 512 U.S. at 644). If so, the law is likely content based; if not, it’s probably content neutral. A law has a content-neutral justification if it “neither references the content of speech . . . nor reflects disagreement with the message such speech conveys.” Id. at 72. As is probably evident, content-neutral laws “pose a less substantial risk of excising certain ideas or viewpoints from the public dialogue” than those that are content based. Id. at 70 (quoting Turner I, 512 U.S. at 642).

A pair of examples illustrate the distinction between content-based and content-neutral speech restrictions. In Brown v. Entertainment Merchants Ass’n, the Supreme Court held that a California law banning vendors from selling “violent video games” to minors was an impermissible content-based speech restriction. 564 U.S. 786, 789 (2011). “Violent” speech, the Court reasoned, is a particular kind of speech—a particular kind of content—and California’s law singled it out for unfavorable treatment vis-à-vis speech that didn’t depict violence. Id. at 799. Citing gory examples from Grimm’s Fairy Tales, The Odyssey, Inferno, and Lord of the Flies, the Court reasoned that violent content has historically been protected for minors. Id. at 795–96. So, regardless of form—whether in print, television, or video games—California’s attempt to draw USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 26 of 59

a distinction between violent and nonviolent speech was impermissibly content based. Id. at 799. Even as to minors, who enjoy more limited First Amendment protections, the government doesn’t possess “a free-floating power to restrict the ideas to which children may be exposed.” Id. at 794 (emphasis added). Accordingly, the Court concluded, speech that is protected for minors—including violent speech—“cannot be suppressed solely to protect the young from ideas or images that a legislative body thinks unsuitable for them.” Id. at 795 (quoting Erznoznik v. City of Jacksonville, 422 U.S. 205, 213 (1975)).

In TikTok, by contrast, the Supreme Court held that a federal statute that required a Chinese social-media company either to sell its U.S.-based operations to U.S. owners or to cease operations in this country was content neutral. TikTok, 604 U.S. at 66, 71. The Court reasoned that the law was facially content neutral because its prohibitions didn’t “target particular speech based upon its content,” “regulate speech based on its function or purpose,” or “impose a ‘restriction, penalty, or burden’ by reason of content” on the company’s service. Id. at 71. Put simply, it didn’t “facially regulate ‘particular speech because of the topic discussed or the idea or message expressed.’” Id. (quoting Reed, 576 U.S. at 163). And, the Court concluded, the statute’s justification was also content neutral: to prevent China, “a foreign adversary[],” “from collecting vast amounts of sensitive data from” millions of American users. Id. at 71–72. That purpose, the Court said, was “unrelated to the content of expression,” and it “neither reference[d] the content of speech USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 27 of 59

on TikTok nor reflect[ed] disagreement with the message such speech convey[ed].” Id. at 72 (citing Ward, 491 U.S. at 792–93). HB³ more closely resembles the law at issue in TikTok than the one in Brown. HB³ is facially content neutral: It regulates social-media platforms based on their functional capabilities and features, not the content they communicate. The law neither references the substance of the speech on any platform nor draws distinctions based on it. Put differently, HB³ applies to platforms wholly without regard to the topics discussed or the ideas or messages conveyed on them. Truth Social is just as likely as Instagram to be affected by HB³ so long as (1) a proportion of minors under 16 spend an average of at least two hours a day on it; (2) it uses infinite scrolling, push notifications, auto-play, or livestreaming, or displays user-engagement metrics; and (3) it selects content algorithmically for users. Fla. Stat. § 501.1736(1)(e). And tellingly, neither a platform nor a user can “avoid or mitigate” the effects of HB³ by “altering” the substance of any posted expression. TikTok, 604 U.S. at 71 (quoting Turner I, 512 U.S. at 644). HB³’s justification is also content neutral: to “protect[] children from [social-media] addiction and associated mental-health harms.” Br. of Appellant at 36. That purpose is “unrelated to the content of expression” on any given platform, and it doesn’t “reflect[] disagreement with the message such speech conveys.” Tik- Tok, 604 U.S. at 72. There’s no indication, for instance, that Florida enacted HB³ because it disagreed with social-media companies’ decisions to “deplatform” users who “advocate for terrorism,” Moody, USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 28 of 59

603 U.S. at 736–37, deprioritize posts on “conspiracy theories about the origin of the [COVID] virus,” Murthy v. Missouri, 603 U.S. 43, 51 (2024), or fact-check “misleading claims about elections,” id. at 50. 8 By contrast, HB³ is easily distinguished from the violentvideo-games restriction at issue in Brown. That law was doomed because it separated video games with violent content from video games without it. HB³ makes no comparable distinction.² CCIA has counterarguments, of course, and they deserve careful consideration.

First, CCIA contends that HB³ abridges its member platforms’ First Amendment rights by restricting “a website’s choices about how to organize and present collections of third-party speech,” which, as already noted, Moody held to be “protected expressive activity.” Br. of Appellees at 38 (citing Moody, 603 U.S. at 731–32). Maybe so, but the fact that a law regulates expressive activity (or more accurately here, “disproportionate[ly] burden[s]” it in a way that we’ve found HB³ does, see supra at 22–23) merely puts

in its exceptions. For instance, the term “social-media platform” is defined not to include “an online service, website, or application where the exclusive function is e-mail or direct messaging consisting of text, photographs, pictures, images, or videos shared only between the sender and the recipients . . . .” Fla. Stat. § 501.1736(1)(e). Whether HB³ applies to an email or messaging service has nothing to do with the subject matter or topic of any correspondence. USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 29 of 59

the First Amendment on the table—it doesn’t ipso facto render the law unconstitutional. The next question in the decision tree—and in practical effect, often the most important—is whether the law regulates based on content. And on that score, HB³ bears scant resemblance to the law at issue in Moody. The Supreme Court treated that law as content based because it restricted social-media platforms’ freedom to “delet[e], alter[], label[], or deprioritiz[e]” certain posts—which is to say, it specifically hindered their ability to curate and moderate their content as they saw fit. Moody, 603 U.S. at 720. Put simply, the law “interfere[d] with [platforms’] editorial choices . . . by ordering the excluded to be included.” Id. at 732. HB³ doesn’t restrict platforms’ content-moderation determinations or tools, but rather it regulates only front-end access and website mechanics (e.g., so-called “addictive features”). Second, CCIA compares HB³ to a hypothetical statute that “restrict[s] all minors from accessing libraries just because some spend (in the state’s view) too much time reading page-turning novels,” or a law that limits minors “from using streaming services because some spend too much time watching engaging cartoons.” Br. of Appellees at 17. Surely, CCIA says, such restrictions would be unconstitutional—and so, the argument goes, must be HB3. Id. Perhaps the laws that CCIA imagines would be unconstitutional— we needn’t say, and so we don’t. But if those laws did violate the First Amendment, it wouldn’t be because they were content based—but rather because, for whatever reason, they failed the USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 30 of 59

scrutiny applicable to content-neutral regulations, a topic we’ll address soon enough. To be clear, though, those laws would be content neutral.

Third, CCIA contends that HB³ is content based because it applies only to so-called “social” speech: The law, CCIA asserts, targets websites “based on the ‘social’ subject matter ‘of the material they disseminate.’” Br. of Appellees at 26 (quoting NetChoice v. Reyes, 748 F. Supp. 3d 1105, 1122–23 (D. Utah 2024)) (citation modified). Dissenting from our order staying the preliminary injunction, Judge Rosenbaum similarly contended that HB³’s restrictions are a “proxy for a content-based regulation directly prohibiting ‘public, interactive, user-to-user social speech.’” Comput. & Commc’ns Indus. Ass’n v. Uthmeier, No. 25-11881, 2025 WL 3458571, at *15 (11th Cir. Nov. 25, 2025) (Rosenbaum, J., dissenting) (emphasis in original). Respectfully, we disagree. The “social”-ness of speech is a function of its form, not its substance. Put simply, “social” speech, as CCIA seems to be using the term, is simply speech that occurs between users on social-media platforms, wholly without regard to “the topic discussed or the idea or message expressed.” TikTok, 604 U.S. at 70. Users can engage in so-called “social” speech about any topic, from the sublime (religion, philosophy, etc.) to the ridiculous (celebrity gossip, cat videos, etc.). Finally, CCIA asserts that HB³’s focus on the “addictive features” of social-media platforms, rather than, say, streaming services, shows that Florida is “plainly focused on particular content.” Br. of Appellees at 16. But that’s not quite right, either. Regardless USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 31 of 59

of whether the content on a social-media platform is a clip from The Bachelor or a political meme, the platform is subject to HB3. It can’t “avoid or mitigate” the effects of HB³ “by altering [its] speech.” TikTok, 604 U.S. at 71. * * * For the foregoing reasons, we conclude that while HB³ restricts First Amendment-protected activity, it likely does so in a content-neutral manner.

C

A content-neutral law like HB³ is subject to intermediate scrutiny, meaning that it is valid so long as it “advances important governmental interests unrelated to the suppression of free speech and does not burden substantially more speech than necessary to further those interests.” TikTok, 604 U.S. at 70 (quoting Turner II, 520 U.S. at 189). A regulation “advances” the government interest so long as it “would be achieved less effectively absent the regulation.” Free Speech Coal., 606 U.S. at 496 (quoting TikTok, 604 U.S. at 70).

A few intermediate-scrutiny corollaries: First, to satisfy intermediate scrutiny—unlike strict scrutiny—a speech regulation need not be the “least restrictive means of serving the State’s interest.” Id. (quoting Ward, 491 U.S. at 798) (citation modified). Even if other “approaches are equally or more effective, under intermediate scrutiny a ‘regulation will not be invalid simply because a court concludes that the government’s interest could be adequately served by some less-speech-restrictive alternative.’” Id. at USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 32 of 59

497–98 (quoting Ward, 491 U.S. at 800). Second, “[t]he regulation’s validity does not turn on our agreement with the legislature concerning the most appropriate method for promoting significant government interests or the degree to which those interests should be promoted.” Id. at 496 (citation modified) (quoting Ward, 491 U.S. at 800). Finally, intermediate scrutiny does not impose on speech regulations a “‘freestanding underinclusiveness limitation,’ and the Government ‘need not address all aspects of a problem in one fell swoop.’” TikTok, 604 U.S. at 76 (quoting Williams-Yulee v. Fla. Bar, 575 U.S. 433, 449 (2015)).

Let’s evaluate HB³ against each of intermediate scrutiny’s two core requirements—again, that it (1) advance an “important governmental interest[] unrelated to the suppression of free speech” and (2) “not burden substantially more speech than necessary” to further that interest. TikTok, 604 U.S. at 70. 1 HB³ easily satisfies the first requirement. CCIA doesn’t dispute that HB³ advances an important state interest; it acknowledges that “protecting minors is certainly a laudable goal.” Br. of Appellees at 28. Even so, it says, HB³ fails intermediate scrutiny’s first element because Florida’s interest “is plainly related to the suppression of expression.” Id. In particular, CCIA emphasizes that HB³ prohibits addictive features only on websites, not on other “services that disseminate speech”—like, for instance, “books that end in cliffhangers.” Id. at 28–29. Florida singles out social-media plat- USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 33 of 59

forms, CCIA insists, because they host speech that Florida “distrusts.” Id. at 17. CCIA’s argument doesn’t necessarily follow. It would hold up only if websites have an inherent expressive value that differs from the expressive value inherent in all books. But both books and websites communicate content that spans the full range of expressive values; they just do so in different formats. Absent a restriction on the substance of the topics, ideas, or messages communicated by websites as compared to those communicated by books, HB³’s “singling out” of websites is not, as CCIA argues, “plainly related to the suppression of expression.”⁹ Br. of Appellees at 28. 2 HB³ also likely satisfies intermediate scrutiny’s second requirement—that it be appropriately tailored to advancing Florida’s interest in protecting minors. See TikTok, 604 U.S. at 70. In partic-

to the extent it seeks to protect minors from “addictive” speech features, HB³ is underinclusive inasmuch as it doesn’t also cover literary devices. But setting aside the question whether literary devices are addictive, even if the state believed them to be an issue, it “need not address all aspects of a problem in one fell swoop” because intermediate scrutiny imposes no “freestanding underinclusiveness limitation.” TikTok, 604 U.S. at 76 (quoting Williams-Yulee, 575 U.S. at 449). Further, the analogy to literary devices that make books more engaging—like cliffhangers, shorter chapters, or serialization—simply reinforces Florida’s contention that it’s not focused on speech and the accompanying ideas, topics, and messages, but rather on features that in and of themselves are agnostic as to content.

USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 34 of 59

ular, HB³ doesn’t “burden substantially more speech than necessary” because it (1) calibrates its restrictions by age, (2) regulates minors’ creation of social-media accounts, not necessarily their access to the platforms themselves, (3) covers only those platforms that have been determined to induce heavy use by children, and (4) applies only to platforms that use both algorithms and one of several addictive features. Id.; see Fla. Stat. § 501.1736(1)(e). First, HB³ is age delimited. It prohibits minors under 14 from creating accounts, id. § 501.1736(2)(a), and it requires 14- and 15-year-olds to obtain parental consent before doing so, id. § 501.1736(3)(a). Notably, it imposes no limitations for those 16 years of age and older. The law thus applies only to younger minors and recognizes the fine-grained distinctions even among those to whom it applies—namely, that 13-year-olds (in general, middleschoolers) are different from 14- and 15-year-olds (those starting high school), who, in turn, are different from those 16 and up (budding adults).

Second, even among those to whom HB³ does apply, the law restricts only their ability to create their own accounts, not their ability to access social-media platforms per se. The creation of a social-media account is a commercial act—a business transaction. Even a minor—indeed, even a young child—may continue to use the platforms provided he either (1) uses his parents’ (or someone else’s) account or (2) gains access without an account. Third, HB³ targets only the platforms to which many minors are already addicted. As a reminder, a platform comes within USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 35 of 59

HB³’s purview only if at least 10% of daily active users under 16 years old spend an average of two hours or more per day on it. Fla. Stat. § 501.1736(1)(e). Together with its other limitations, HB³’s tailoring efforts fall within the broad range of potential legislative solutions that intermediate scrutiny permits. And fourth, HB³ operates not as an outright prohibition of platform usage but rather as a “conditional ban” of sorts—namely, on platforms’ use of addictive features and algorithms. See TikTok, 604 U.S. at 74, 77. The law applies only to platforms that use a combination of algorithms and addictive features and that succeed in keeping many minors on them for an average of at least two hours a day. See Fla. Stat. § 501.1736(1)(e). Refraining from using either algorithms or addictive features, or remaining below the temporal threshold, exempts a platform from the law’s application. Now, none of that is necessarily to say that HB³ embodies the best policy choice. Maybe, maybe not. But intermediate scrutiny affords governments “latitude . . . to design regulatory solutions to address content-neutral interests.” TikTok, 604 U.S. at 77 (quoting Turner II, 520 U.S. at 213). Accordingly, it is within Florida’s discretion to draw the lines it has—many of which, of course, narrow the law’s scope. So, for instance, Florida may validly opt to prohibit minors under 14, rather than 16, from creating social-media accounts, and to require 16-year-olds, rather than all those under 18, to obtain their parents’ consent before creating accounts. So too, it may validly train its attention on platforms that a segment of children use more than two hours per day, rather than one. And USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 36 of 59

finally, it is free to focus on sites that use both algorithms and “addictive features,” rather than one or the other of those. The numerous limitations on HB³’s sweep—that it applies only to users under the age of 16, operates differently on 14- and 15-year-olds than on younger children, restricts only account creation and not access per se, and affects only those platforms using features most likely to addict minors—differentiate it from the law at issue in Packingham v. North Carolina, 582 U.S. 98 (2017), on which CCIA relies. There, applying intermediate scrutiny, the Supreme Court invalidated a North Carolina statute making it a felony for any registered sex offender to access any “commercial social networking Web site” that allowed any minors to use it. Id. at 101. That law operated with “extraordinary breadth.” Id. at 109 (Alito, J., concurring in the judgment). Unlike HB³, which applies only to a subset of minors, the statute there covered all registered sex offenders. And unlike HB³, which applies only to a subset of social-media platforms, the law there defined the covered sites exceedingly broadly; it included “commonplace social media websites like Facebook and Twitter,” as well as “websites as varied as Amazon.com, Washingtonpost.com, and Webmd.com.” Id. at 101, 106 (majority opinion). And unlike HB³, which restricts account creation but doesn’t limit access to the platforms per se, the statute there forbade any usage of the covered sites. Id. at 101. HB³ is tailored in a way that the law at issue in Packingham was not. USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 37 of 59

CCIA complains that Florida failed to adopt social-media restrictions that would have restricted less speech. Citing the Supreme Court’s decision in an abortion-clinic-buffer-zone case, McCullen v. Coakley, 573 U.S. 464 (2014), CCIA contends that the state didn’t “demonstrate that alternative measures that burden substantially less speech would fail to achieve the government’s interests.” Br. of Appellees at 32 (quoting McCullen, 573 U.S. at 495). One such alternative, CCIA says, would have been to defer the decision of whether children can open social-media accounts entirely to their parents, who “already have many tools to protect their minors on the Internet.” Id.

But again, as the Supreme Court recently reiterated in Free Speech Coalition, under intermediate scrutiny—at least as conventionally applied, and as applied in Internet-related cases—a regulation needn’t be the “least restrictive means of serving the State’s interest.” 606 U.S. at 496 (quoting Ward, 491 U.S. at 798) (citation modified). Even if other “approaches are equally or more effective,” HB³ isn’t “‘invalid simply because a court concludes that the government’s interest could be adequately served by some lessspeech-restrictive alternative.’” Id. at 497–98 (quoting Ward, 491 U.S. at 800). Instead, the legislature has some discretion to choose “the most appropriate method for promoting” its interest in curbing minors’ social-media addiction or “the degree to which th[at] interest[] should be promoted.” Id. at 496 (quoting Ward, 491 U.S. at 800). Florida chose to prohibit the creation of social-media accounts for those under 14 because it found that relying exclusively USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 38 of 59

on parental supervision and consent was “ineffective.” Br. of Appellant at 8; see also States’ Amicus Br. at 7 (“Parental controls have already proven insufficient to stem the tide of harm kids are suffering. And besides, kids inevitably find workarounds.”).¹⁰ CCIA seeks to distinguish Free Speech Coalition on the ground that the law at issue there was less speech-restrictive than HB3. In that case, the Supreme Court upheld under intermediate scrutiny a Texas statute requiring pornographic websites to verify users’

looks like a least-restrictive-means requirement, even in the context of intermediate scrutiny. There, in the course of invalidating a Massachusetts statute that made it a crime to stand on a public street or sidewalk within 35 feet of an abortion clinic, the Supreme Court said that, for intermediate-scrutiny purposes, the law burdened substantially more speech than necessary because the state hadn’t “seriously undert[aken] to address the problem with less intrusive tools readily available to it.” McCullen, 573 U.S. at 494. The Court faulted Massachusetts for failing to use other tools at its disposal—such as existing general criminal statutes prohibiting assault, harassment, or breach of the peace—to “ensur[e] public safety outside abortion clinics” and “prevent[] harassment and intimidation of patients and clinic staff.” Id. at 490, 492; see also Fla. Preborn Rescue, Inc. v. City of Clearwater, 161 F. 4th 732, 745 (11th Cir. 2025) (applying McCullen to hold that a city ordinance prohibiting people from standing on a public sidewalk five feet away from an abortion-clinic driveway burdened substantially more speech than necessary because the city could have enforced the state’s existing anti-obstruction statute to achieve the same interest).

The simple response to McCullen—and its arguable imposition of a least-restrictive-means-like requirement as part of an intermediate-scrutiny analysis—is that Free Speech Coalition is both (1) more recent and (2) as an Internet-related case, more closely on point. USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 39 of 59

ages. 606 U.S. at 495. In the course of so holding, the Court observed that the Texas law prohibited minors from accessing speech—namely, pornography—that was unprotected as to them. Id. CCIA contends that, by contrast, HB³ prohibits certain minors from accessing protected speech. That may be so, but Free Speech Coalition didn’t purport to establish a constitutional outer limit; nothing in the Court’s reasoning there precludes a state from imposing an age-verification requirement that operates to limit minors’ access to some protected speech—provided, of course, that it does so in a content-neutral manner.

To the contrary, everything the Court said in Free Speech Coalition seems to point in the other direction. The Court emphasized, for instance, that “[r]equiring age verification is common when a law draws lines based on age.” Id. at 479. It noted that many states require proof of age to buy alcohol and tobacco, as well as to obtain licenses to carry a handgun, to drive, and to marry. Id. And in a similar vein, it observed that brick-and-mortar stores have long engaged in the “widespread” practice of verifying a purchaser’s age in person for “age-restricted services, such as online gambling, alcohol and tobacco sales, and car rentals.” Id. at 479, 497. And, importantly, the Court stressed that in the online world, “the need for age verification . . . is even greater” because “[u]nlike a store clerk, a website operator cannot look at its visitors and estimate their ages.” Id. at 481. * * * USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 40 of 59

For the foregoing reasons, we hold that HB³ is likely constitutional as applied to CCIA’s members. Yes, it implicates the First Amendment and thus triggers constitutional scrutiny. But it likely does so in a content-neutral manner, and it likely survives intermediate scrutiny because it doesn’t burden substantially more speech than necessary to advance the important interest in protecting minors from social-media addiction.

IV

Last but not least, we must separately assess CCIA’s facial challenge in light of the Supreme Court’s recent decision in Moody v. NetChoice, LLC, 603 U.S. 707 (2024). As we understand it, CCIA’s facial challenge takes two forms: what we’ll call a “traditional” facial challenge, in which it seeks to vindicate its members’ First Amendment rights; and a facial “overbreadth” challenge, in which it seeks to vindicate its members’ users’ rights. For reasons previously explained, we decline to examine the user-focused overbreadth challenge, as CCIA seems (at least for now) to have walked away from it. See supra at 17 n.4. That leaves the more traditional, member-focused facial challenge.

As the district court observed, CCIA raised both facial and as-applied First Amendment challenges to HB3. And despite some initial uncertainty, the parties seem to agree that the district court considered and adjudicated both claims. See Supp. Br. of Appellant at 20–21; Supp. Br. of Appellees at 19–20. For instance, in granting CCIA’s preliminary-injunction motion, the district court held that CCIA was “substantially likely to succeed not just in [its] as-applied USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 41 of 59

challenge but in [its] facial challenge as well.” Dist. Ct. Order, June 3, 2025, at 52–53 (emphasis added).

We hold that CCIA has not carried its facial-challenge “burden” under Moody. 603 U.S. at 744. As the Supreme Court warned there, the decision to mount a facial challenge “comes at a cost.” Id. at 723. Because “facial challenges threaten to short circuit the democratic process,” they are, by design, “hard to win.” Id. (citation modified); see also id. at 745 (Barrett, J., concurring) (“[D]ealing with a broad swath of varied platforms and functions in a facial challenge strikes me as a daunting, if not impossible, task.”); id. at 748 (Jackson, J., concurring in part and concurring in the judgment) (“[P]laintiffs bringing a facial challenge must clear a high bar.”); id. at 778 (Alito, J., concurring in the judgment) (“[P]arties mounting facial attacks [must] satisfy demanding requirements.”). Even in First Amendment cases, where the facial-challenge bar is lower than it otherwise might be, the plaintiff must show that “a substantial number of [the law’s] applications are unconstitutional, judged in relation to the statute’s plainly legitimate sweep.” Id. at 723 (majority opinion) (quoting Ams. for Prosperity Found. v. Bonta, 594 U.S. 595, 615 (2021)).

To conduct the required Moody analysis, a reviewing court proceeds in two steps. First, it must “assess the state laws’ scope,” asking, “What activities, by what actors, do the laws prohibit or otherwise regulate?” Id. at 724. Having done that, the court must then “decide which of the laws’ applications violate the First Amendment, and . . . measure them against the rest.” Id. at 725. USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 42 of 59

To succeed on a facial challenge, a plaintiff must show that “the law’s unconstitutional applications substantially outweigh its constitutional ones.” Id. at 724.

CCIA asserts that HB³ is invalid on its face because it “is unconstitutional in all its applications.” Br. of Appellees at 43 n.7 (emphasis added). Echoing the district court, CCIA reasons that HB³ “applies to any service where people upload content and view the content of others” and thus “implicates the First Amendment in every application except for a service that only has unconstitutional speech.” Oral Arg. 28:44–28:56. And Florida, CCIA says, hasn’t shown “that there’s any service that’s covered by this law that is all about only unconstitutional speech.” Id. 28:57–29:03; accord Dist. Ct. Order, June 3, 2025, at 27–28 (observing that HB³ covers only social media platforms “on which protected speech . . . proliferate[s]” and that Florida hadn’t “hypothesize[d]” that the law could cover sites whose “content is all or mostly unprotected speech”). Respectfully, we don’t think that’s quite right. To be sure, HB³ applies to websites that enable users to “‘upload content or view the content or activity of other users’—in other words, engage in speech.” Dist. Ct. Order, June 3, 2025, at 43 n.25 (citing Fla. Stat. § 501.1736(1)(e)). But those features—uploading and viewing content—aren’t necessarily limited to websites that host only fully protected speech. Put differently, just because users are “engag[ing] in speech” on a platform doesn’t mean that their expression is fully protected by the First Amendment. Id. There are all kinds of social-media platforms featuring—or even if not featuring, USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 43 of 59

permitting—all kinds of expression. To be sure, much of that expression is fully protected. But some isn’t. The Supreme Court has held, for instance, that certain kinds of speech enjoy no First Amendment protection at all—e.g., incitement, true threats. See Brandenburg v. Ohio, 395 U.S. 444 (1969) (incitement); Counterman v. Colorado, 600 U.S. 66 (2023) (true threats). Other kinds of speech enjoy constitutional protection when exhibited to adults but not when exhibited to children—pornography being the most obvious example. See Free Speech Coal., 606 U.S. at 496. It may well be true that most expression on most social-media platforms is indeed protected. But not all of it is, at least not with respect to all users. So, for instance, imagine a platform that hosts only unprotected speech—say, Incitement.com or True- Threats.com. A law restricting children’s (or anyone’s) access to it wouldn’t implicate the First Amendment at all and would trigger only rational-basis review. A law restricting access to a site that hosts speech that is protected for adults but not children—like pornography—would implicate the First Amendment, but a contentneutral restriction—like an age-verification requirement—would be subject only to intermediate scrutiny. See Free Speech Coal., 606 U.S. at 495. Accordingly, even if HB³ “imposes the same onerous restriction” on every platform, Dist. Ct. Order, June 3, 2025, at 43, that restriction could well be subject to different levels of scrutiny, potentially leading to different outcomes.

USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 44 of 59

It’s no answer to say, as both CCIA and the district court have, that Florida has failed to demonstrate constitutional applications of HB3. See Oral Arg. 24:20–24:32 (“Under the First Amendment, it’s the state’s burden to prove that the law has more constitutional applications than unconstitutional applications.”); Dist. Ct. Order, June 3, 2025, at 28. In a facial-challenge posture, we don’t think it’s the state’s obligation to do so. Rather, the Supreme Court seems to have put the “burden” squarely on the plaintiff: As one “price” of its “decision to challenge the law[] as a whole,” the challenger must “carry its burden” of demonstrating that the invalid applications substantially outweigh the valid ones. Moody, 603 U.S. at 744.

One might interpose a “come on” objection at this point: Surely, the argument would go, the overwhelming majority of platforms to which HB³ applies overwhelmingly host protected speech. And if, as CCIA insists, the law’s application to those websites would be unconstitutional, then the law’s unconstitutional applications would probably substantially outweigh the constitutional ones. But however intuitive, that sort of impressionistic judgment won’t satisfy Moody’s stringent demands. Whether or not it’s likely that the collective content of the platforms within HB³’s ambit is overwhelmingly protected, CCIA bears the “burden” of demonstrating both (1) the universe of the law’s applications and (2) that the invalid applications substantially outweigh the valid ones. See id.

USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 45 of 59

To discharge its facial-challenge burden, CCIA must provide the reviewing court with the data—either hard facts or, at the very least, sufficient briefing—necessary to conduct the Moody-required comparative analysis. It hasn’t done so here. Indeed, CCIA’s briefing doesn’t mention even a single platform that isn’t one of its members—let alone any nonmember platform that (in whole or in part) hosts unprotected speech. In that respect, we think, CCIA makes the same mistake that it and NetChoice made in Moody: treating the case “more like [an] as-applied claim[] than like [a] facial one[].” Moody, 603 U.S. at 724; cf. NetChoice, LLC v. Bonta, 152 F. 4th 1002, 1020 (9th Cir. 2025) (“[F]acial challengers cannot focus on a law’s effect on their own platforms—they cannot even limit the inquiry to the tech giants.”); NetChoice, LLC v. Yost, 180 F. 4th 268, 312 (6th Cir. 2026) (Batchelder, J., concurring in the judgment) (refraining from addressing NetChoice’s facial challenge “without a developed record”).

We’ve learned the facial-challenge lesson the hard way. In Moody, the Supreme Court, even while largely agreeing with our merits analysis, faulted us for “confining [our] analysis” to “certain heartland applications” instead of “address[ing] the full range of activities the law[] cover[ed].” Moody, 603 U.S. at 724; see NetChoice, LLC v. Att’y Gen., Fla., 34 F. 4th 1196 (11th Cir. 2022), abrogated by Moody, 603 U.S. at 724. The Court went on to observe that even if it were to attempt the facial analysis itself, it couldn’t “undertake the needed inquiries” because the “parties ha[d] not briefed the critical issues [t]here, and the record [wa]s underdeveloped.” 603 U.S. at 726. The Court therefore vacated and remanded our case (and USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 46 of 59

its companion, NetChoice, L.L.C. v. Paxton, 49 F. 4th 439 (5th Cir. 2022), abrogated by Moody, 603 U.S. at 724), “to consider the scope of the laws’ applications, and weigh the unconstitutional as against the constitutional ones.” Id. at 726. In turn, both we and our colleagues on the Fifth Circuit remanded our respective cases to the district courts for further development of the issues pertinent to the facial-challenge analysis. See Order, NetChoice, LLC v. Att’y Gen., State of Fla., No. 21-12355, (11th Cir. Aug. 16, 2024), Dkt. No. 178; NetChoice, L.L.C. v. Paxton, 121 F. 4th 494, 500 (5th Cir. 2024) (remanding to the district court to “develop a factual record” in response to Moody).

We are similarly stymied here. For the foregoing reasons, we conclude that CCIA has failed to satisfy Moody because it hasn’t adequately addressed the “full range” of HB³’s applications, nor does any part of the record bear on the constitutionality of those applications.

V

To recap: We hold (1) that CCIA has associational standing to assert the First Amendment rights of its social-media platform members; (2) that the district court didn’t abuse its discretion in declining to abstain under Younger; and (3) that although HB³ implicates First Amendment interests, it likely passes constitutional muster, as it is a content-neutral speech restriction that satisfies intermediate scrutiny. Finally, (4) we reject the facial challenge to USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 47 of 59

HB³ on the ground that CCIA hasn’t carried its burden of demonstrating, as required by Moody, that the law’s unconstitutional applications substantially outweigh its valid applications. Accordingly, we AFFIRM the district court’s determinations that CCIA has associational standing and its decision not to abstain under Younger, REVERSE the preliminary injunction with respect to CCIA’s First Amendment claim as applied to its members, VACATE the preliminary injunction with respect to CCIA’s facial First Amendment challenge, and REMAND to the district court for proceedings consistent with this opinion.

USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 48 of 59

NEWSOM, Circuit Judge, concurring: Needless to say, I join the Court’s opinion in full. I write separately to flag an issue that implicates CCIA’s as-applied challenge and that arises from (of all places) the Supreme Court’s recent decision in West Virginia v. B.P.J. ex rel. Jackson, 146 S. Ct. 2356 (2026). There, in the course of holding that states may, consistently with the Equal Protection Clause, limit participation in women’s and girls’ sports to biological females, the Court sort of said—well no, expressly said, in so many words—that “as-applied challenges are not entertained when courts apply intermediate scrutiny or rational basis review.” Id. at 2377 n.5. Now, to be sure, for that proposition, the Court cited another equal-protection case. See id. (citing Tuan Anh Nguyen v. INS, 533 U.S. 53, 70 (2001)). But just as surely, the Court emphasized that “the intermediate scrutiny test . . . applies in a variety of contexts” and, for the stock formulation of the standard, it cited a prominent First Amendment decision. See id. at 2376 (citing Ward v. Rock Against Racism, 491 U.S. 781, 801 (1989)).

So, the question: Is B.P.J. properly read to foreclose as-applied challenges for all claims—including First Amendment claims—that trigger intermediate scrutiny? The answer doesn’t so much matter here—because, for reasons explained in the main opinion, we’ve concluded that CCIA’s as-applied challenge fails “on the merits,” so to speak, on the ground that HB³ survives a proper intermediate-scrutiny analysis. See Maj. Op. at 31–40. But if B.P.J. were read for all it might be worth, perhaps we could have USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 49 of 59

gotten away with a much shorter opinion. And in any event, the answer may very well matter in the next First Amendment case, in which an as-applied plaintiff could otherwise show that a law would fail intermediate scrutiny.

For reasons I’ll explain, I don’t think we (yet) have a clear (enough) indication from the Supreme Court that it actually meant in B.P.J. to cut off all as-applied challenges to laws reviewable under an intermediate-scrutiny framework. But I suppose it might have. In any event, I thought the question worth noting.

I

I should begin by briefly unpacking B.P.J. Again, the Supreme Court held there that a state doesn’t “violate[] the Equal Protection Clause by maintaining female sports teams for biological females.” 146 S. Ct. at 2373. The Court began by articulating its topline premises and conclusion: (1) Laws that segregate sports teams based on sex are “sex-based classification[s] that trigger[] intermediate scrutiny”; (2) the sex-based classification that such laws draw is “substantially related” to the states’ “important” interests in “safety and competitive fairness”; and therefore, (3) “schools may determine eligibility for women’s and girls’ sports based on biological sex.” Id. at 2374–75. Straightforward enough. The Court then shifted gears to address the plaintiffs’ “narrower” argument—namely, that even if the states’ interests in safety and competitive fairness are “sufficient reasons to exclude most biological males,” they “do not justify excluding [the] relatively small subclass of biological males . . . who identify as female USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 50 of 59

and have taken puberty blockers or hormones.” Id. at 2375 (second emphasis in original). The Court observed that the plaintiffs “characterize[d] their equal protection argument” in several “overlapping ways,” two of which are relevant for our purposes. Id. First, the plaintiffs asserted “a lack of substantial relationship between the laws’ biological-sex classification” and the states’ asserted interests. Id. And second, the plaintiffs’ framed their attack “as an asapplied challenge to the biological-sex classification.” Id. The Court rejected both approaches.

With respect to the plaintiffs’ first tack, the Court emphasized that its “equal protection precedents allow general [sexbased] classifications like those” in the laws before it “so long as there is at least a substantial relationship between the classification and the State’s interests.” Id. at 2376. Reiterating its earlier conclusion that “‘[p]erfection in making the necessary classifications is neither possible nor necessary,’” id. (quoting Mass. Bd. of Ret. v. Murgia, 427 U.S. 307, 314 (1976)), the Court observed that “none of [its] ‘gender-based classification equal protection cases have required that the statue under consideration must be capable of achieving its ultimate objective in every instance,’” id. (quoting Nguyen, 533 U.S. at 70). In further support of its explanation, the Court looked beyond the equal-protection context to First Amendment cases, noting that “under the intermediate scrutiny test that . . . applies in a variety of contexts,” the “‘validity of the regulation depends on the relation it bears to the overall problem the government seeks to correct, not on the extent to which it furthers the government’s interests in an individual case.’” Id. (quoting USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 51 of 59

Ward, 491 U.S. at 801); see also id. (citing United States v. Edge Broadcasting Co., 509 U.S. 418, 430–31 (1993), for the proposition that the intermediate-scrutiny analysis is focused on the relationship between the restriction and “the general problem” rather than “an individual case”). Accordingly, the Court held that “States are not required to conduct an individual-by-individual comparison of the physical and athletic capabilities of all biological males in order to satisfy intermediate scrutiny.” Id. at 2377. The Supreme Court also rebuffed the plaintiffs’ second framing of their position—namely, “as an as-applied equal protection argument.” Id. The Court found this second argument to be “basically the same as the[ir] first” one. Id. As the Court described the theory, the plaintiffs contended that even if the states’ sex-based classification was “generally permissible,” it violated the Equal Protection Clause “as applied to those biological males such as [the plaintiffs,] who identify as female and have taken puberty blockers or hormones.” Id. The Court held, to the contrary, that “as long as the relationship” between a state’s interest and the classification “is sufficient as a general matter, the State is not constitutionally required to grant individualized exemptions to specific athletes or subclasses.” Id. It was at this point in its analysis that the Court dropped footnote five, in which it said the thing: Distinguishing its earlier decisions in Caban v. Mohammed, 441 U.S. 380 (1979), and Lehr v. Robertson, 463 U.S. 248 (1983), the Court emphasized that a more recent decision had “clarified that as-applied challenges are not entertained when courts apply intermediate scrutiny or rational basis re- USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 52 of 59

view.” B.P.J., 146 S. Ct. at 2377 n.5 (emphasis added) (citing Nguyen, 533 U.S. at 70). Back up above the line, the Court then rejected “[t]he plaintiffs’ as-applied argument [as] in essence a backdoor argument for strict scrutiny, which requires a much tighter relationship between the State’s classification and the State’s asserted interests.” Id. at 2378.

II

So, what to make of B.P.J.—and in particular the (potentially) portentous footnote five? Did the Court there really mean to foreclose all as-applied challenges to laws that (like HB³) trigger intermediate scrutiny, including those brought under the First Amendment? Candidly, I’m not quite sure. I can see arguments going both ways.

A

Here’s the case, as I see it, for what I’ll call the “broad” reading. First, and (by far) most significant, is the starkness of the Court’s chosen language. Once again, as relevant here: “[A]s-applied challenges are not entertained when courts apply intermediate scrutiny.” Id. at 2377 n.5. Second, and in a similar vein, the Court referred to the intermediate-scrutiny standard generally, emphasizing that it “applies in a variety of contexts.” Id. Third, as already noted, when explaining what it took to be the proper application of that “test,” the Court looked, not to an equal-protection case, but rather to a pair of First Amendment decisions: “[T]he ‘validity of the regulation depends on the relation it bears to the overall problem the government seeks to correct, not on the extent to USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 53 of 59

which it furthers the government’s interests in an individual case.’” Id. at 2376 (quoting Ward, 491 U.S. at 801, and citing Edge Broadcasting, 509 U.S. at 430–31).

Finally, and perhaps most fundamentally, the logic underlying the B.P.J. Court’s ultimate rejection of the plaintiffs’ contention would seem to apply just as surely to First Amendment cases as to equal-protection cases. The “as-applied argument” that the plaintiffs there were making, the Court warned, was “in essence a backdoor argument for strict scrutiny, which requires a much tighter relationship between the State’s classification and the State’s asserted interest.” Id. at 2378; see also Adams v. Sch. Bd. of St. Johns Cnty., 3 F. 4th 1299, 1329 (11th Cir. 2021) (Pryor, C.J., dissenting) (requiring the government to prove tailoring “as to each plaintiff who brings an as-applied challenge would disregard intermediate scrutiny by demanding a perfect fit between the sex-based classification and the government interest at issue.”), rev’d en banc, 57 F. 4th 791 (11th Cir. 2022). That critique is at least arguably generalizable across contexts because, by their very nature, as-applied challenges exist in some tension with less rigorous forms of scrutiny. On the one hand, plaintiff-by-plaintiff analysis is of course the very sine qua non of as-applied litigation. On the other, intermediate scrutiny (to say nothing of rational-basis review) “affords the Government latitude in designing a regulatory solution,” Turner Broad. Sys., Inc. v. FCC, 520 U.S. 180, 213 (1997), and, famously, does not require a perfect fit between ends and means, see Free Speech Coal. v. Paxton, 606 U.S. 461, 471 (2025). But—and this is the point—that “latitude” will essentially evanesce if the government USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 54 of 59

has to prove that its interests apply to each and every successive individual who happens to file suit.

It’s perhaps particularly easy to see that dynamic at work in a context like sex-segregated sports, where an insistence that the government repeatedly justify wholesale distinctions case by case at the retail level would risk swallowing those distinctions whole. Though biological men are on average more athletically capable than women, “[n]ot every biological male athlete is bigger, stronger, faster, or otherwise more athletically able than every biological female athlete.” B.P.J., 146 S. Ct. at 2376 (emphases in original). But if every less capable male athlete could seek an as-applied exemption from general rules requiring sex-segregated sports, the distinction between men’s and women’s sports would die a death of a thousand cuts. The same phenomenon, though, may well exist in the First Amendment context. Imagine, for instance, a contentneutral ordinance prohibiting large late-night gatherings in a suburban park. On its face, the restriction would probably survive intermediate scrutiny as a sufficiently narrowly tailored means of preventing after-dark disruptions. Cf. Ward, 491 U.S. at 796. But as-applied challenges introduce a wrinkle: Must the city repeatedly prove—in every successive case—that each prospective gathering is likely to be so loud that it would disturb neighborhood peace? If it turns out that the city can’t always do so, then the as-applied exemptions will make Swiss cheese out of an otherwise facially valid noise-prevention ordinance.

USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 55 of 59

Bottom line: It may just be that if “the intermediate scrutiny test that the Court applies in a variety of contexts”—equal protection, free-speech, etc.—is indeed focused on the relationship between the law at issue and “‘the overall problem the government seeks to correct, not on the extent to which it furthers the government’s interests in an individual case,’” B.P.J., 146 S. Ct. at 2376 (quoting Ward, 491 U.S. at 801), then it’s a test built largely—if not entirely—for facial challenges.

B

There is, though, another side to the story. Here’s the case, as I see it, for the “narrow” reading of B.P.J. First, and most obviously, there’s the fact that the Court was deciding an equal-protection case. And at times, it seemed to frame its analysis specifically in those terms. See, e.g., id. at 2376 (“[T]his Court’s equal protection precedents allow general classifications like those made in the [laws before us] so long as there is at least a substantial relationship between the classification and the State’s interests.”); id. (“[N]one of this Court’s ‘gender-based classification equal protection cases have required that the statute under consideration must be capable of achieving its ultimate objective in every instance.’”); id. at 2377 (“Intermediate scrutiny permits a sex-based classification that, as here, is ‘not invidious, but rather realistically reflects the fact that the sexes are not similarly situated in certain circumstances.’”).¹

1 Indeed, there are suggestions in the Court’s opinion that it might have meant

to focus its analysis even more narrowly on equal-protection cases that arise in the sports context. See, e.g., B.P.J., 146 S. Ct. at 2376 (“the sports context USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 56 of 59

Second, although the Court cited its First Amendment decisions in Ward and Edge Broadcasting for support, it did so only in addressing the first of the plaintiffs’ two “narrower” arguments— namely, to say that the plaintiffs’ just misunderstood the level of generality at which a proper intermediate-scrutiny analysis should be conducted. See id. at 2376. The Court didn’t (at least explicitly) circle back to those cases in the course of rejecting what it called the plaintiffs’ second, “as-applied” argument. See id. at 2377. One final reason for circumspection: Recall that, in support of what is for our purposes the key statement—that “as-applied challenges are not entertained when courts apply intermediate scrutiny,” the Court cited its earlier decision in Nguyen. Id. at 2377 n.5. Not only is Nguyen an equal-protection case, but I’m not sure I read it to say everything that the Court in B.P.J. seemed to attribute to it. I certainly don’t find there anything that purports to expressly foreclose as-applied challenges for all claims that trigger intermediate scrutiny.

III

So, what here? This case, of course, presents an as-applied challenge to a content-neutral speech regulation that triggers intermediate scrutiny. See Maj. Op. at 31. Accordingly, if B.P.J.’s rule is indeed generalizable—if, in fact, “as-applied challenges are not entertained when courts apply intermediate scrutiny or rational basis

is . . . crucial”); id. at 2377 (“the distinctive sports context”); id. at 2378 (“[e]specially in the sports context”); id. (“[p]articularly in the sports context”). USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 57 of 59

review,” full stop, 146 S. Ct. at 2356 n.5—then CCIA’s claim failed the moment that we concluded that intermediate scrutiny applied. I’m reluctant to go there—at least yet.

First, there’s precedent and consistent practice—a lot of it. We have long (and seemingly without qualm) entertained as-applied First Amendment challenges to laws that trigger intermediate scrutiny. See, e.g., United States v. Romain, --- F. 4th ----, 2026 WL 2250001, at *4 (11th Cir. Aug. 5, 2026); Moms for Liberty v. Brevard Pub. Schs., 118 F. 4th 1324, 1336–37 (11th Cir. 2024); Fort Lauderdale Food Not Bombs v. City of Fort Lauderdale, 11 F. 4th 1266, 1291–92 (11th Cir. 2021); cf. Ranch House, Inc. v. Amerson, 238 F. 3d 1273, 1284–85 (11th Cir. 2001) (remanding to the district court to consider an as-applied First Amendment challenge under intermediate scrutiny). As have our sister circuits, in cases too numerous to count. See, e.g., Green v. U.S. Dep’t of Just., 54 F. 4th 738, 745–46 (D.C. Cir. 2024); Brokamp v. James, 66 F. 4th 374, 397–403 (2d Cir. 2023); 360 Virtual Drone Servs. LLC v. Ritter, 102 F. 4th 263, 278–80 (4th Cir. 2024); Hines v. Pardue, 117 F. 4th 769, 779–85 (5th Cir. 2024); People for the Ethical Treatment of Animals, Inc. v. Reynolds, 173 F. 4th 959, 966–69 (8th Cir. 2026); Project Veritas v. Schmidt, 125 F. 4th 929, 952–60 (9th Cir. 2025) (en banc). I’m hesitant to conclude that the B.P.J. Court abrogated all of that law and convention without a clearer indication that it really meant to extend its reasoning to the First Amendment context. And I’d be doubly hesitant given USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 58 of 59

that, just last year, in TikTok Inc. v. Garland, the Supreme Court itself entertained an as-applied First Amendment challenge while applying intermediate scrutiny. See 604 U.S. 56, 74–80 (2025). Second, there’s Moody v. NetChoice, LLC, 603 U.S. 707 (2024). That decision, too, it seems to me, stands in some tension with a prohibition on as-applied First Amendment challenges to laws that trigger intermediate scrutiny. The Supreme Court was crystal clear there: As-applied challenges are the rule, and facial challenges the rare, “hard to win” exception. Id. at 723. And for good reason. Facial challenges “‘threaten to short circuit the democratic process’ by preventing duly enacted laws from being implemented in constitutional ways.” Id. (quoting Wash. State Grange v. Wash. State Republican Party, 552 U.S. 442, 451 (2008)). But if the Supreme Court’s point in Moody was to make as-applied challenges the new normal in First Amendment litigation, it seems strange that the same Court—the same nine people—would have then turned around less than two years later and nixed as-applied challenges for the whole range of First Amendment claims that call for intermediate scrutiny.² And I don’t think it makes me a result-oriented hack to worry that a combination of Moody’s new (or newly emphasized)

2 See, e.g., Ward, 491 U.S. at 791 (content-neutral, time, place, or manner re-

strictions); United States v. O’Brien, 391 U.S. 367, 376 (1968) (expressive conduct); Cent. Hudson Gas & Elec. Corp. v. Pub. Serv. Comm’n, 447 U.S. 557, 562– 65 (1980) (commercial speech); City of Renton v. Playtime Theatres, Inc., 475 U.S. 41, 47–49 (1986) (regulation of secondary effects); Turner Broad., 520 U.S. at 186 (content-neutral regulation of the airwaves); Paxton, 606 U.S. at 478 (regulations that only incidentally burden protected speech). USCA11 Case: 25-11881 Document: 94-1 Date Filed: 10/14/2026 Page: 59 of 59

limitations on facial First Amendment challenges and a prohibition on as-applied challenges in a host of contexts would lead to, well, a lot less protection for traditionally safeguarded speech. * * * In the end, I’ll confess that I’m not quite sure what to make of B.P.J.’s statement casting doubt on the viability of as-applied challenges to laws that trigger less than strict scrutiny. But before exporting it to the First Amendment context, I think it’s wise to await a clearer signal from the Supreme Court.