Free Speech Coal., Inc. v. Paxton

Before Higginbotham, Smith, and Elrod, Circuit Judges. Jerry E. Smith, Circuit Judge: Texas H.B. 1181[1] was scheduled to go into effect on September 1, 2023. It imposes new standards on commercial pornographic websites, requiring them to verify the age of their visitors and to display health warnings about the effects of the consumption of pornography. Free Speech Coalition, Incorporated, an adult industry trade association; several domestic and foreign corporations that produce, sell, and host pornography; and one individual adult content creator brought a facial challenge against the enforcement of H.B. 1181. The day before the law was scheduled to take effect, the district court granted plaintiffs’ motion for a preliminary injunction in an 81-page order. Free Speech Coal., Inc. v. Colmenero, No. 1:23-CV-917, 2023 WL 5655712, at *30 (W.D. Tex. Aug. 31, 2023). The court concluded that: (1) both the age-verification requirement and the health warnings of H.B. 1181 likely violate plaintiffs’ First Amendment rights, and (2) as to certain plaintiffs, the law likely conflicts with, and thus is preempted by, Section 230 of the Communications Decency Act, 47 U.S.C. § 230. Then the court ruled that plaintiffs had satisfied the other preliminary injunction factors. Texas filed an emergency appeal, and this court issued an administrative stay, ordered expedited briefing, and heard oral argument two weeks later. A month after argument, this panel granted Texas’s motion to stay the district court’s injunction pending appeal. We now vacate that stay and rule on the merits of the preliminary injunction. First, we vacate the injunction against the age-verification requirement based on Ginsberg v. New York, 390 U.S. 629 (1968), which remains binding law, even after Ashcroft v. ACLU (Ashcroft II), 542 U.S. 656 (2004).[2] The proper standard of review is rational-basis, not strict scrutiny. Applying rational-basis review, the age-verification requirement is rationally related to the government’s legitimate interest in preventing minors’ access to pornography. Therefore, the age-verification requirement does not violate the First Amendment. Further, Section 230 does not preempt H.B. 1181. So, the district court erred by enjoining the age-verification requirement. Second, we affirm the injunction in regard to the health warnings. The district court properly applied National Institute of Family & Life Advocates v. Becerra (NIFLA), 585 U.S. 755, 766 (2018), and ruled that H.B. 1181 unconstitutionally compelled plaintiffs’ speech. I. H.B. 1181 regulates only certain entities, specifically, “commercial entit[ies] that knowingly and intentionally publish[] or distribute[] material on an Internet website, including a social media platform, more than one-third of which is sexual material harmful to minors.” § 129B.002(a). Those regulated entities must take two actions. First, they must “use reasonable age verification methods” to limit their material to adults. Id. Second, they must “display notices on the landing page of the website and on all advertisements for that website in 14-point font or larger.” § 129B.004(1) (cleaned up). The newly enacted statute defines sexual material harmful to minors by adding “with respect to minors” or “for minors,” where relevant, to the well-established Miller test for obscenity. See Miller v. California, 413 U.S. 15, 24 (1973).[3] It also mimics the language of 47 U.S.C. § 231, which the Supreme Court reviewed in Ashcroft II.[4] Regulated entities may choose their preferred “reasonable age verifi-cation methods,” including by outsourcing the process to a third party. §§ 129B.002–129B.003. The options include the use of government-issued identification. See § 129B.003(b)(2)(A). But the websites may alternatively require digital identification or may use other “commercially reasonable method[s].” § 129B.003(b)(1), (2)(B). Further, whoever performs the verification may not retain any of the individual’s identifying information. § 129B.002(b). The websites must also show three health warnings on their landing pages and advertisements. § 129B.004(1).[5] Additionally, the entities must place a notice at the bottom of every webpage. § 129B.004(2).[6] Should an entity either refuse or fail to comply with H.B. 1181, the Attorney General may seek injunctive relief and civil penalties of up to $10,000 for each day a company lacks age-verification; up to $10,000 for each instance of improper retention of identifying information; and up to $250,000 for a minor’s accessing of sexual material harmful to minors. § 129B.006.[7] Shortly after Texas enacted H.B. 1181 and before it took effect, plain-tiffs sued. They claimed, inter alia, H.B. 1181 impermissibly encroaches on their First Amendment rights and, for some plaintiffs, conflicts with Section 230. The district court found that: (1) all plaintiffs have standing and that sovereign immunity does not bar the claims because Ex parte Young creates a carveout for suits against state officials where the plaintiffs seek prospective relief for violation of constitutional rights[8]; (2) the age-verification requirement is subject to and fails strict scrutiny under Ashcroft II and Reno[9]; (3) the health warnings compel speech, so they are subject to, and fail, strict scrutiny; and (4) Section 230 conflicts with and therefore preempts H.B. 1181 as to certain plaintiffs. Thus, the district court believed that plaintiffs were likely to succeed on the merits of their claim and to suffer irreparable harm and that the balance of harms and public interest favored a preliminary injunction under Winter v. NRDC, Inc., 555 U.S. 7, 20 (2008). So, the court issued a pre-enforcement preliminary injunction. Texas now appeals. II. We review preliminary injunctions for abuse of discretion. Ashcroft II, 542 U.S. at 664. But such “a decision grounded in erroneous legal principles is reviewed de novo.” Mock v. Garland, 75 F.4th 563, 577 (5th Cir. 2023) (quoting Byrum v. Landreth, 566 F.3d 442, 445 (5th Cir. 2009)). Preliminary injunctions are “extraordinary remed[ies] that may only be awarded upon a clear showing that the plaintiff is entitled to such relief.”[10] The moving party must show (1) a substantial likelihood of success on the merits, (2) a substantial threat of irreparable injury if the injunction is not issued, (3) that the threatened injury if the injunction is denied outweighs any harm that will result if the injunction is granted, and (4) that the grant of an injunction will not disserve the public interest. Mock, 75 F.3d at 577 (quoting Byrum, 566 F.3d at 445). And “[t]he govern-ment’s and the public’s interests merge when the government is a party.” Id. (citing Nken v. Holder, 556 U.S. 418, 435 (2009)). The key issue is whether the district court properly found a substantial likelihood of success on the merits. So, we turn to that first. III. A. H.B. 1181′s age-verification requirements are subject to rational-basis review. Applying that standard, we uphold them as constitutional. 1. “The State has an interest to protect the welfare of children and to see that they are safeguarded from abuses.” Ginsberg, 390 U.S. at 640 (cleaned up).[11] For that reason, regulations of the distribution to minors of materials obscene for minors are subject only to rational-basis review. See id. at 641; see also Ent. Merchs., 564 U.S. at 793–94 (quoting Ginsberg, 390 U.S. at 641). Ginsberg dealt with a First Amendment challenge to a New York law criminalizing the sale of so-called “girlie” picture magazines and applied the then-relevant Memoirs obscenity standard, modified for children.[12] The Court recognized that the magazines at issue “are not obscene for adults” but ruled that, because the seller could still stock and sell them to adults, Butler v. Michigan did not apply. Ginsberg, 390 U.S. at 634–35 (citing Butler, 352 U.S. 380, 382–84 (1957) (holding that the state cannot “reduce the adult population of Michigan to reading only what is fit for children”)). Instead, New York could criminalize the selling of those magazines to children because “it was not irrational for the legislature to find that exposure to material condemned by the state is harmful to minors.” Id. at 641. The decision in Erznoznik v. City of Jacksonville, 422 U.S. 205 (1975), reaffirmed a robust reading of Ginsberg‘s principle: “It is well settled that a State or municipality can adopt more stringent controls on communicative materials available to youths than on those available to adults.” Id. at 212 (citing Ginsberg, 390 U.S. 629). Crucially, the material regulated by the ordinance in Erznoznik was available to both youths and adults.[13] “Assuming the ordinance [was] aimed at prohibiting youths from viewing the films,” the Court did not take issue with the burdens that enforcing the regulation would have on adults. Id. at 213. Instead, the Court questioned the regulation’s targeting of material, noting that “it sweepingly forbids display of all films containing any uncovered buttocks or breasts, irrespective of context or pervasiveness. Thus it would bar a film containing a picture of a baby’s buttocks . . . .” Id. But that is no issue here; H.B. 1181 is restricted to material obscene for minors. Erznoznik suggests that if—like H.B. 1181—Jacksonville’s ordinance had been tailored to material obscene for minors, the Court stood ready to accept Jacksonville’s contention that “the . . . ordinance [was] a reasonable means of protecting minors from this type of visual influence.” See id. at 212. Ginsberg‘s central holding—that regulation of the distribution to minors of speech obscene for minors is subject only to rational-basis review— is good law and binds this court today. And not only this court. Years after Reno and Ashcroft II, the Supreme Court, to avoid rational-basis review, felt required to distinguish Ginsberg in Entertainment Merchants, 564 U.S. at 794. As that Court described it, Ginsberg “approved a prohibition on the sale to minors of sexual material that would be obscene from the perspective of a child” because that proscription “‘was not irrational.’” Id. at 793–94 (quoting Ginsberg, 390 U.S. at 641). But the material at issue in Entertainment Merchants was violent, not sexual—and Ginsberg set out a lower standard only for regulation of certain kinds of sexual content. Further, the Court and its individual Justices have cited Ginsberg multiple other times after Reno and Ashcroft II, albeit for different propositions.[14] In an attempt to distinguish Ginsberg, Plaintiffs pick at the factual dissimilarities between the world of Ginsberg and our world. They note that “the Supreme Court recognized that source-based restrictions on Internet expression raise concerns categorically different from those at issue in cases such as Ginsberg, given the nature of cyberspace and the breadth and invasiveness of speech burdens in this context.”[15] Also, plaintiffs posit that in-person age-verification creates fewer risks to privacy because “[m]any adults are never even asked for their identification in person; their appearance alone suffices.”[16] We disagree with those analyses. First, as plaintiffs admit, the statute at issue in Ginsberg necessarily implicated, and intruded upon, the privacy of those adults seeking to purchase “girlie magazines.” But the Court still applied rational-basis scrutiny. Second, the age-verification requirements do not impose any sort of “categorically different” burden on adults. H.B. 1181 provides that a purveyor of pornography can use “digital verification,” “government-issued identification,” or other “commercially reasonable method[s].” § 129B.003(b). That allows for at least three concrete means of age-verification: (1) government ID, (2) facial appearance, or (3) some other available information used to infer the user’s age. At least one of those options will have no more impact on privacy than will in-person age verification à la Ginsberg. Moreover, H.B. 1181 punishes entities $10,000 for each instance of retention of identifying information, possibly yielding heavier penalties than would the failure to age-verify.[17] Third, even were there a gap in privacy concerns as large as plaintiffs suggest, we decline to adopt their notion that such a gap matters. In short, no binding precedent compels us to depart from Ginsberg on privacy grounds, and we decline to do so. Finally, the Supreme Court itself has declined to adopt such a distinction. If the differences between the contemporary world of the Internet and the 1960′s world of in-person interaction were sufficient to distinguish Gins-berg, the Court would have noted as much in Reno. Yet none of the four “important respects” the Court notes distinguishing the statute in Reno from the statute in Ginsberg references the Internet at all. See Reno, 521 U.S. at 865–66.[18] Instead of following Ginsberg, plaintiffs point to two cases that address laws like H.B. 1181 but that applied strict scrutiny: Reno and Ashcroft II. The former is readily distinguishable. There, the Court reviewed and enjoined the Communications Decency Act of 1996 (“CDA”). Reno, 521 U.S. at 849. In doing so, it recognized four “important respects” in which the CDA and the statute in Ginsberg differed. Id. at 865. Most of those same differences exist between the CDA and H.B. 1181—but those are not the only differences between the two laws; they include the following: (1) The CDA included prohibitions on non-sexual material; H.B. 1181 does not. Id. at 873.[19] (2) Parental participation or consent could not circumvent the CDA; it can circumvent H.B. 1181. Id. at 865. (3) The CDA did not specifically define the pro-scribed material; H.B. 1181 does. Id. at 873. (4) The CDA had no limitation to commercial activity; H.B. 1181 covers only commercial entities. Id. at 865.[20] (5) In enjoining the CDA, the Court relied at least in part on “the absence of a viable age verification process,” but that process is the central requirement of H.B. 1181. Id. at 876. Finally, (6) the Court’s decision was fundamentally bound up in the rudimentary “existing” technology of twenty-seven years ago, but technology has dramatically developed. Id. at 876–77. Indeed, only one distinction appears to work against our analysis here. The Court noted that “the New York statute defined a minor as a person under the age of 17, whereas the CDA, in applying to all those under 18 years, includes an additional year of those nearest majority.” Id. at 865–66. Given that this is the sole mention of that distinction, and that the rest of the distinctions still align with H.B. 1181, it beggars belief that the Supreme Court meant that to be an essential component in triggering the Ginsberg frame-work. That is especially true where the Court relied principally on the CDA’s overbreadth and lack of adherence to the Miller standard. See id. at 873.[21] Nor does other seemingly contradictory language in Reno impede our analysis. For example, Reno says that the interest in protecting children, for which it cites Ginsberg, “does not justify an unnecessarily broad suppression of speech addressed to adults.” Id. at 875. As noted above, the reach of the CDA was well beyond Ginsberg‘s safe harbor and included even non-sexual material.[22] Moreover, Reno makes that point, referring to Sable Communications, Inc. v. FCC, 492 U.S. 115 (1989). See Reno, 521 U.S. at 875. Sable addressed “an outright ban on indecent as well as obscene interstate commercial telephone messages.” 492 U.S. at 117. That too is well outside Gins-berg‘s safe harbor for regulations on distribution to minors of material obscene for minors.[23] On the other hand, Ashcroft II supplies plaintiffs’ best ammunition against H.B. 1181. After all, despite Texas’s protestations, H.B. 1181 is very similar to COPA. Sure, COPA was criminal, and H.B. 1181 is civil. And COPA allowed age-verification as an affirmative defense, yet H.B. 1181 requires it upfront. But those changes do not affect our analyses here.[24] Ash-croft II, finding that COPA probably failed the narrow tailoring component of strict scrutiny, sent the case back down for trial. 542 U.S. at 673. One might read Ashcroft II for the proposition that COPA (and consequently H.B. 1181) fail strict scrutiny. We can even assume that here.[25] But that assumption does not end our analysis. Though Ashcroft II concluded that COPA would fail strict scrutiny, it contains startling omissions. Why no discussion of rational-basis review under Ginsberg? And why no analysis of intermediate scrutiny under Renton v. Playtime Theatres, Inc., 475 U.S. 41 (1986)? We find those omissions particularly surprising considering that the Court in Reno felt the need to distinguish those at length. See Reno, 521 U.S. at 865–68. We see only one answer and therefore only one way to read Ashcroft II consistently with Ginsberg: Ashcroft II did not rule on the appropriate tier of scrutiny for COPA. It merely ruled on the issue the parties presented: whether COPA would survive strict scrutiny. Indeed, the petitioner’s brief in Ashcroft II made two claims: I. COPA is narrowly tailored to further the government’s compelling interest in protecting minors from harmful material on the World Wide Web. . . . II. The court of appeals erred in holding that COPA is not narrowly tailored. Brief for Pet’r, Ashcroft II, 542 U.S. 656 (2004) (No. 03-218), 2003 WL 22970843, at *iii. In other words, the petitioners did not challenge the applicable standard of review. Because that is not a jurisdictional argument, the Court did not have to correct them sua sponte.[26] It is true that, in passing and without reference to Ginsberg, the Court said, “[w]hen plaintiffs challenge a content-based speech restriction, the Government has the burden to prove that the proposed alternatives will not be as effective as the challenged statute.” Ashcroft II, 542 U.S. at 657. But that is the closest that opinion comes to ruling on the appropriate standard of review.[27] Given our circuit’s respect for the Court’s dicta, we cannot brush that comment aside as such; yet, it is inapposite for two reasons: First, read in context, it is most readily seen as an explanation of how strict scrutiny works generally and not as a clear articulation of its appropriateness for COPA. Second, and more importantly, it is inconsistent with the proposition that Ginsberg remains good law.[28] Ginsberg undeniably upholds a content-based restriction on speech under a rational-basis framework. But reading that passage from Ashcroft II in a vacuum is irreconcilable with that proposition. Because the Court makes clear that Ginsberg is good law after Ashcroft II, Ginsberg‘s on-point frame-work must take pride of place. See Ent. Merchs., 564 U.S. at 793. In short, the question of the appropriate standard of review in Ashcroft is a “[q]uestion[] which merely lurk[s] in the record, neither brought to the attention of the court nor ruled upon” and consequently is not “to be considered as having been so decided as to constitute precedent[].” Cooper Indus., Inc. v. Aviall Servs. Inc., 543 U.S. 157, 170 (2004) (quoting Webster v. Fall, 266 U.S. 507, 511 (1925)).[29] Ashcroft II does not control. Plaintiffs point to Playboy to counter Texas’s suggestion that Reno and Ashcroft do not supply the standard of review here. Playboy dealt with a stat-ute that “required cable television operators who provide channels primarily dedicated to sexually-oriented programming either to fully scramble or otherwise fully block those channels or to limit their transmission to hours when children are unlikely to be viewing.” 529 U.S. at 806 (cleaned up). To plaintiffs’ credit, Playboy seems to have the clearest language supplying a standard of review: “As we consider a content-based regulation, the answer should be clear: The standard is strict scrutiny.” Id. at 814. Moreover, Play-boy dealt with a law the purpose of which was to protect children from inci-dentally consuming pornographic material. See id. at 806. Further, even after acknowledging the difference in “degree,” the Court held that “content-based burdens must satisfy the same rigorous scrutiny as . . . content-based bans.” Id. at 812. But Playboy cannot surmount the rock that is Ginsberg. As we have discussed, Ginsberg is good law.[30] So it must have some minimum content. H.B. 1181 is plainly more like the regulation in Ginsberg than like the regulation in Playboy. H.B. 1181 allows adults to access as much pornography as they want whenever they want. The law in Playboy did not. The burden in Playboy, although not a ban, is different in kind from whatever “burden” arises from the same type of age-verification required to enter a strip club, drink a beer, or buy cigarettes. The law in Ginsberg, like H.B. 1181, targeted distribution to minors; the law in Playboy targeted distribution to all. That is, once certain an individual is not a minor, H.B. 1181 does nothing further.[31] The same cannot be said of the law in Playboy, which imposed substantial burdens even after an individual established his or her majority. Moreover, Playboy preceded not only Entertainment Merchants but also Reno. If we should read Playboy as broadly as plaintiffs suggest, it renders Reno‘s distinguishing of Ginsberg inexplicable.[32] See Reno, 521 U.S. at 864– 66. Likewise, broadcast media has always raised medium-specific consider-ations that meaningfully diminish the guidance that First Amendment cases concerning broadcast media can provide in other technological contexts— such as regulations to combat “signal bleed,” Playboy, 529 U.S. at 806.[33] Finally, an appeal to H.B. 1181′s “content-based” nature is insufficient because the law in Ginsberg was content-based.[34] Ginsberg carves out an exception to heightened scrutiny of content-based speech restrictions. If Ginsberg is no longer good law, we await that instruction, but until that day it must stand for something. We agree with the dissent’s laudable conclusion that “Texas has the right—and . . . the obligation—to protect its minors, and in doing so, it must have the means to frustrate their access to pornographic materials consistent with the First Amendment.” But the dissent’s read of Reno, Ashcroft, Play-boy, Sable, and Ginsberg would strangle the state’s ability to do just that. Because it is never obvious whether an internet user is an adult or a child, any attempt to identify the user will implicate adults in some way. In the dissent’s view, any attempt to protect children will be subject to strict scrutiny, often a death knell in and of itself. [35] To suggest protecting children would beso difficult is inconsistent with Ginsberg, where rational basisreview was sufficient even though adults would presumably have to identify them-selves to buy girlie magazines. In other words, the dissent’s reading implies that the invention of the Internet somehow reduced the scope of the state’s ability to protect children. That is a dubious principle without support in existing Supreme Court caselaw.[36] Plaintiffs proffer a few more reasons why strict scrutiny applies. None convinces. First, they emphasize that the law does not just regulate speech obscene to minors. Instead, because of the “one-third threshold,” the act regulates even “content . . . benign for people of any age.” Plaintiffs assert that means that “a substantial number of its applications are unconstitutional,” so the statute is facially unconstitutional. United States v. Stevens, 559 U.S. 460, 473 (2010) (internal quotation marks and citation omitted). That averment fails for two reasons. First, it very well might be appropriate in this context to evaluate plaintiffs’ websites as a whole.[37] And second, the magazines at issue in Ginsberg were similarly situated to the websites captured by H.B. 1181. Indeed, the “girlie magazines” of Ginsberg‘s day had a substantial amount of content that was non-sexual in its entirety.[38] The inclusion of some—or even much—content that is not obscene for minors does not end-run Ginsberg where the target of the regulation contains a substantial amount of content that is obscene for minors. Second, plaintiffs assert that “the Act’s vagueness further chills protected speech.” In particular, they contend that the phrase “with respect to minors” has no fixed meaning.[39] That was not a problem for the law in Gins-berg, and it is no problem here. Third, plaintiffs maintain that strict scrutiny applies because “the Act discriminates based on speaker and viewpoint.” They point to both what they deem H.B. 1181′s “underinclusiveness” and its health-warnings requirement as evidence that Texas is really engaged in speaker discrimination “to stigmatize the ‘porn industry’ and deter all patronage of such disfavored speech.” The under-inclusivity in question is the exemption of “search engines . . . and social-media platforms . . . that display the same content.” On this, there is no issue. First, under-inclusivity (in the speaker/-viewpoint discrimination context) only serves as a signal that the state may be engaged in viewpoint discrimination. See, e.g., City of Ladue v. Gilleo, 512 U.S. 43, 52 (1994) (under-inclusivity “diminish[es] the credibility of the government’s rationale for restricting speech in the first place”). Where, as here, the exemptions are driven by something else, though—such as making a reasonable policy choice to avoid the legal concerns that accompany at-tempts to regulate the “entire universe of cyberspace,” Reno, 521 U.S. at 868—the state is not pursuing viewpoint discrimination. Second, under-inclusivity is typically not an issue where the state chooses to regulate a specific medium. See, e.g., Gilleo, 512 U.S. at 52 (discussing “[e]xemptions from an otherwise legitimate regulation of a medium of speech” (emphasis added)). And plaintiffs’ examples fall into line with this reasoning.[40] Plaintiffs suggest that under R.A.V. v. City of St. Paul, under-inclusivity “alone [is] enough to render the ordinance presumptively invalid[.]” 505 U.S. 377, 394 (1992). But the sort of “[s]electivity” at issue in R.A.V. was between different sorts of messages, not different mediums. See id. In short, where a state chooses to regulate a specific kind of medium, that selection does not necessarily implicate strict scrutiny based on viewpoint discrimination.[41] Thus, bound by Ginsberg and the Supreme Court’s application of it in Entertainment Merchants and Erznoznik, we apply rational-basis review. 2. Under rational-basis review, H.B. 1181 easily surmounts plaintiffs’ constitutional challenge. As the Court in Ginsberg puts it, we need “only . . . be able to say that it was not irrational for the legislature to find that exposure to material condemned by the statute is harmful to minors.” 390 U.S. at 641. We do that easily. The record is replete with examples of the sort of damage that access to pornography does to children. One study finds that earlier use of adult pornography was correlated with an increased likelihood of engagement “with deviant pornography (bestiality or child).” A review of literature from 2013–2018 finds a correlation between “frequent use of online pornography” and “distorted gender orientations, insecurities and dissatisfaction about one’s own body image, depression symptoms, assimilation to aggressive models,” and more. Another review of scientific literature finds that “internet pornography addiction fits into the addiction framework and shares similar basic mechanisms with substance addiction.” Further, a study finds that “[t]he more boys used sexually explicit internet content, the poorer their school grades were six months later.” That is far more than what is necessary to demonstrate that the legislature did not act irrationally.[42] Thus, H.B. 1181′s age-verification requirement likely passes constitutional muster under the rational-basis standard in Ginsberg.[43] We offer no opinion as to how it would fare under any other standard of review. B. We turn to plaintiffs’ likelihood of success in challenging the health warnings. Concluding that the health warnings unconstitutionally compel speech, we uphold the injunction in that regard only. H.B. 1181 regulates only commercial entities. It is well established that the First Amendment protects commercial speakers.[44] Further, it pro-tects both the “right to speak and the right to refrain from speaking.”[45] Yet, it is equally settled that the government may regulate commercial speech more heavily than non-commercial speech.[46] So, the constitutionality of the required warnings turns on both (1) whether the speech is commercial and (2) the applicable level of scrutiny. As with the age-verification requirements, the parties dispute the standard of review. Plaintiffs claim the warnings are content-based regulations of non-commercial speech—so strict scrutiny applies.[47] Texas responds by asserting that the warnings compel only commercial speech, so we should review deferentially under Zauderer.[48] Alternatively, the state suggests that if Zauderer does not apply, this court should use Central Hudson‘s intermediate scrutiny.[49] 1. We answer first whether H.B. 1181 regulates commercial or non- commercial speech. The district court offered several reasons that this speech is not commercial. First, it found that plaintiffs’ speech goes “beyond proposing a commercial transaction.” 2023 WL 5655712, at *21 (cleaned up) (quoting Bolger v. Youngs Drug Prods. Corp., 463 U.S. 60, 66 (1983)). Second, it ruled that this speech is not commercial because the “paid access that makes speech commercially viable is ‘inextricably intertwined’ with the speech itself.” Id. at *21 (quoting Riley, 487 U.S. at 796). But both justifycations misconstrue the nature of these websites. Though courts have not settled “the precise bounds of the category of expression that may be termed commercial speech, . . . it is clear enough that [some of] the speech at issue in this case—advertising pure and simple— falls within those bounds.” Zauderer, 471 U.S. at 637.[50] That suffices for the part of the statute that covers advertisements. Along with warnings on advertisements, though, H.B. 1181 also requires warnings on the landing page and on each subsequent page of the regulated websites. So, we must resolve whether those webpages also propose commercial transactions.[51] As always, we begin with the text of the statute. H.B. 1181 regulates “commercial entit[ies]” and defines such as “ a corporation, limited liability company, partnership, limited partnership, sole proprietorship, or other legally recognized business entity.” § 129B.001(1).[52] The text’s focus on business and corporate forms reads most naturally as an indication that the statute reaches only those entities that publish or distribute sexual material harmful to minors for commercial or business purposes. Thus, the landing page and other pages of those subscription-based and paid websites are proposing “no more than” a commercial transaction—e.g., “you give us money and we give you porn.”[53] Indeed, to the degree that the websites purport to offer educational speech; they only offer it as an add-on to their primary purpose, exchanging far-from-core First Amendment entertainment for money.[54] Therefore, the webpages communicate commercial speech.[55] We also conclude that H.B. 1181 regulates commercial speech on free websites because they too propose a commercial transaction: They offer pornography in exchange for data; then, they monetize that data, primarily through advertisements. That series of transactions is explicitly commercial, just as was the case in NetChoice. See 49 F.4th at 485–88.[56] Finally, we reject the contention that Riley‘s inextricably intertwined” analysis controls here. That the websites might offer non-obscene sexual education or expressive obscenity “no more convert[s]” the act of selling pornography and bartering for data into protected speech than does “ teach-ing home economics . . . convert[] [Tupperware parties] into educational speech.”[57] Indeed, instead of being closer to a paywall on a newspaper— core, protected speech—the landing pages are more like the entrance to a strip club—commercial activity with a speech element.[58] Therefore, we review these speech regulations under the commercial speech doctrine. 2. We next consider whether the law qualifies for Zauderer‘s relaxed scrutiny. In Zauderer, the Court upheld lawyer advertising regulations because the requirements mandated lawyers “include in [their] advertising purely factual and uncontroversial information about the terms under which [their] services will be available.” 471 U.S. at 651. The Court reasoned that a lawyer’s “constitutionally protected interest in not providing any particular factual information in his advertising is minimal” because “disclosure requirements trench much more narrowly on an advertiser’s interests than do flat prohibitions on speech . . . .” Id. Recent decisions have distilled that language to apply Zauderer scrutiny where a state compels “commercial enter-prises to disclose purely factual and uncontroversial information about their services . . . .” Chamber of Comm., 85 F.4th at 768 (quotation marks and citations omitted).[59] In other words, Texas must show that the warnings are both (1) purely factual and (2) uncontroversial. Because the state has not met its burden on the uncontroversial nature of the warnings on the record before us, Zauderer is inapplicable.[60] Assuming the statements are factual, a compelled statement is “un-controversial” for purposes of Zauderer where the truth of the statement is not subject to good-faith scientific or evidentiary dispute and where the statement is not an integral part of a live, contentious political or moral debate.[61] That standard does not mean that whenever the compelled speaker dislikes or disagrees with the message he must convey, the statement is controversial. Nor does it mean that that controversy exists solely because an expert can be found who disagrees with a predominant view. It means only that there must be some widespread, good-faith dispute over the topic or the facts. We need not determine the outer limits of what establishes “controversy” because Texas has failed to rebut plaintiffs’ challenges in such a way that we are comfortably within its boundaries. Thus, Zauderer is inapplicable. We supply two of the many examples of the dueling experts and studies[62]: First, Texas cites a study finding a negative correlative relationship between (1) time spent watching porn and (2) gray-matter volume and brain function in 21-to-45-year-old men. Second, the state’s experts describe the “host of mental health afflictions” that they link to viewing pornography. But plaintiffs respond with similarly credentialed and persuasive experts who, on review of “the last several decades of research,” find “no generally accepted, peer-reviewed research studies or scientific evidence which indicate that viewing adult-oriented erotic material causes physical, neurological, or psychological damage such as ‘weakened brain function’ or ‘impaired brain development.’” And the plaintiffs’ experts refute Texas’s first point—which casts that correlative relationship as causal—by proffering that, to the degree there are causal findings, they typically “run the opposite way of Texas’s claims, i.e., porn viewing is an effect, not a cause of mental issues.” We are not scientific journal editors, much less social scientists, be-havioral experts, or neurologists. The courts generally are not the place to hash out scientific debate, particularly not on so contentious a topic as the impacts of engaging with pornography. Experts must do that in academic journals, studies, and presentations. Therefore, the record leaves us with no option but to declare that the health impacts of pornography are currently too contentious and controversial to receive Zauderer scrutiny. See NIFLA, 585 U.S. at 769.[63] 3. It is unsettled precisely which standard of scrutiny applies to com pelled commercial speech that is not subject to Zauderer scrutiny. On the one hand, Central Hudson applied a form of intermediate scrutiny. On the other, Central Hudson dealt only with restrictions on commercial speech, not com-pelled speech. Yet, NIFLA suggests that compelled speech must survive, at minimum, intermediate scrutiny. See 585 U.S. at 773. So, because H.B. 1181 fails Central Hudson‘s intermediate scrutiny, we need not decide that issue here.[64] Under Central Hudson, courts apply “a four-part analysis” to govern-ment-compelled restriction of commercial speech. 447 U.S. at 566. First, the court “must determine whether the expression concerns lawful activity and is not misleading.” Id. (cleaned up). Second, the court asks “whether the asserted governmental interest is substantial. If both inquiries yield positive answers, we must determine whether the regulation advances the governmental interest asserted, and whether it is not more extensive than is necessary to serve that interest.” Id.; see also City of Cincinnati v. Discovery Net-work, Inc., 507 U.S. 410, 416 (1993). That standard is stricter than rational-basis, but it is not a least-restrictive-means test. Instead, we ask whether it is a reasonable fit, “one whose scope is in proportion to the interest served.” Fox, 492 U.S. at 480 (internal quotation marks and citation omitted). The “government goal [must] be substantial, and the cost . . . carefully calculated.” Id. And the governmental entity bears the burden of proving both the interest and the fit. Because we address here only the application of H.B. 1181 to speech deemed (1) not obscene for adults but (2) obscene for minors, the expression at issue is lawful, meeting the first prong of Central Hudson. Next, we agree that Texas has a substantial—and even compelling[65]—interest in preventing minors from accessing pornography, meeting the second prong.[66] But Texas fails the third and fourth prongs. The health warnings, taken as a whole, might advance the state’s stated interests. The warnings declare the potential harm of minors’ engaging with pornography, and they do so in a noticeable fashion—in a way likely to discourage minors from using and adults from allowing their children to use the websites. But Texas must meet a higher standard than “might.” “[A] governmental body seeking to sustain a restriction on commercial speech must demonstrate that . . . its restrictions will in fact alleviate [the harms] to a material degree.”[67] Because Texas has not made such a showing, we adopt the approach recently taken by the Ninth Circuit: “[C]ompelling sellers to warn consumers of a potential ‘risk’ never confirmed by any regulatory body—or of a hazard not ‘known’ to more than a small subset of the scientific community—does not directly advance” the government’s inter-est. Nat’l Ass’n of Wheat Growers v. Bonta, 85 F.4th 1263, 1283 (9th Cir. 2023). Therefore, Texas fails to satisfy the third prong. Further, as in NIFLA, Texas does not adequately tailor the warnings to the interest. As the district court noted, because of the age-verification requirements, those warnings displayed on the landing page and subsequent pages will presumably not reach any minors. To the degree that minors do see the warnings, the warnings’ language seems beyond the average child’s reading comprehension ability: They use multisyllabic, scientific words and phrases such as “biologically addictive,” “desensitizes brain reward circuits,” and “conditioned responses.” Therefore, the warnings are too broad reasonably to fit the interest.[68] Further, even if scientific findings supported the warnings, but see supra note 63, Texas has made no showing that they will discourage minors who have circumvented the age restrictions from accessing pornography. Additionally, the inclusion of a helpline for “mental health or substance use disorders” cannot be connected directly to preventing children from viewing pornography. § 129B.004(2).[69] Finally, Texas has not made any showing that it tried a government-funded public information campaign or that such a campaign would be ineffective. In other words, we see “numerous and obvious less-burdensome alternatives” to the health warnings. Went For It, 515 U.S. at 633 (quoting Discovery Network, 507 U.S. at 417 n.13). Thus, Texas fails the fourth prong of Central Hudson, making the health warnings unconstitutionally compelled commercial speech. C. Finally, Plaintiffs contend that 47 U.S.C. § 230(c) preempts H.B. 1181. We conclude that it does not. We start with the text of Section 230(c). Congress entitled Subsection (c) “Protection for ‘Good Samaritan’ blocking and screening of offensive material.” Thus, Texas correctly suggests that Congress enacted § 230 to shield against liability for removal, but not promulgation, of “offensive material.” Indeed, the text of § 230(c)(2) makes that clear.[70] Both provisions protect “providers” of “interactive computer service[s]” from liability stemming from attempts to “ restrict” unwanted material. Id. Subsection 230(c)(1), however, lacks that one-way language.[71] But, on its face, it still does not protect plaintiffs for two reasons. First, the act’s context clarifies the subsection’s open-ended language. Cf. Yates v. United States, 574 U.S. 528, 531–32 (2015) (plurality) (A fish is obviously a “tangible object,” but it is not in the context of the Sarbanes-Oxley Act.). Indeed, Texas convincingly suggests that (c)(1) was meant to abrogate the holding in Stratton Oakmont, Inc. v. Prodigy Services Co., 1995 WL 323710 (N.Y. Sup. Ct. May 24, 1995). That court held an online service provider trying to filter out profane content liable for the remaining defamatory contents on a financial bulletin board.[72] But plaintiffs seek to turn that meaning on its head and make (c)(1) a shield for purposefully putting “offensive material” onto the Internet. Second, particularly in light of that context, making plaintiffs liable under H.B. 1181 would not be “treat[ing them] as the publisher or speaker” of the underlying content. 47 U.S.C. § 230(c)(1). This is where Doe v. MySpace, Inc., 528 F.3d 413 (5th Cir. 2008), complicates the analysis. There we held that § 230 shielded My Space from negligence liability for publishing communications between a minor and an adult who later sexually assaulted her. See id. at 417–18. But that case is readily distinct. In MySpace, “[p]arties complain[ed] that they were harmed by a Web site’s publication of user-generated content.” Id. at 419. And that is the point of Section 230: to immunize web service providers for harm caused by unremoved speech on their website. Like cellphone service providers, interactive computer service providers cannot be held liable for harmful communications that they fail to remove. But liability under H.B. 1181 is not like liability under a negligence claim. It is not reliant on the harm done by third-party content. It imposes liability purely based on whether plaintiffs comply with the statute, independently of whether the third-party speech that plain-tiffs host harms anybody. As Texas puts it “if a minor circumvents age verification, ignores the health warnings, and is subsequently harmed by third-party content or resulting offline conduct,” Section 230 would kick in and bar liability. That is the nature of Section 230′s protections: to protect a provider from speaker-liability stemming from the speech it hosts. Liability under H.B. 1181 is plainly different. Plaintiffs reject that harm-based theory based on a misreading of MySpace. Their contention fails for two reasons. First, it relies on a misrepresentation of the court’s characterization of the Third Circuit’s holding in Green v. AOL, 318 F.3d 465 (3d Cir. 2003). Green reasoned that Section 230 barred the following tort claim: “AOL was negligent in promulgating harmful content and in failing to address certain harmful content on its network.” Id. at 471 (emphasis added). Of course, that parallels the claim in MySpace, but it is not analogous here, see supra. Second, even if we accepted plaintiffs’ generous characterization of our court’s characterization of the Green court’s “recharacterization” of the Green plaintiffs’ claims as applicable here, plain-tiffs still do not meet their own standard: “If recharacterizing the plaintiff’s claims shows that the theory of liability is based, e.g., on decisions relating to the monitoring, screening, and deletion of content, the claims are barred and preempted.” Brief for Appellee at 54 (emphasis added, internal citations and quotations omitted). But liability here is not based on “monitoring, screen-ing, [or][73] deletion of content.” Instead, it is based on a failure to age-verify or to include health warnings. Plaintiffs continue by quoting MySpace, 528 F.3d at 420, to claim that the Act “force[s] upon [plaintiffs] a responsibility ‘quintessentially related to a publisher’s role,’ to filter their audience . . . .” But publishers do not filter audiences; they filter content. And in Green, the court says exactly that. “[A]ctions quintessentially related to a publisher’s role” refers to “monitor-ing, screening, and deletion of content.” MySpace, 528 F.3d at 420 (quoting Green, 318 F.3d at 471). Plaintiffs mislead the reader. Plaintiffs make several other claims as to why MySpace controls. First, they cite our statement in MySpace that “Congress provided broad immunity under the CDA to Web-based service providers for all claims stemming from their publication of information created by third parties . . . .” 523 F.3d at 418 (emphasis added by plaintiffs). That misses the point. The emphasis, properly placed, would read, “Congress provided broad immunity under the CDA to Web-based service providers for all claims stemming from their publication of information created by third parties . . . .” Id. (emphasis added). H.B. 1181 is not the sort of liability “stemming from their publication” of pornography that Section 230 immunizes. In fact, plaintiffs’ reliance on that passage is belied immediately by the next quote they select: “Parties complaining that they were harmed by a Web site’s publication of user-generated content . . . may sue the third-party user who generated the content.” Id. at 419 (emphasis added). This underscores the importance of the harm analysis above. Finally, plaintiffs also note that, in MySpace we mentioned in passing MySpace’s failure to provide “age verification software.” See id. at 422. But that does not impact the analysis. It merely means that Section 230 shields from liability interactive computer service providers that do not age verify despite some putative obligation arising from tort law. Therefore, Section 230 does not preempt H.B. 1181. IV. We conclude with the remaining equitable factors for preliminary injunctions. Although plaintiffs are unlikely to succeed on the merits of their challenge against H.B. 1181′s age-verification requirement, they are likely to succeed against the health warnings. So, we turn to those. The remaining factors favor the preliminary injunction against the health warnings. Texas challenges plaintiffs’ assertion of irreparable harm, claiming first that the foreign plaintiffs have no constitutional rights that Texas could infringe and second that plaintiffs have not shown H.B. 1181 will negatively impact viewership. Plaintiffs respond by pointing out that “[t]he loss of First Amendment freedoms, for even minimal periods of time, un-questionably constitute irreparable injury.” Elrod v. Burns, 427 U.S. 347, 373 (1976) (plurality opinion) (citing N.Y. Times Co. v. United States, 403 U.S. 713 (1971)).[74] Plaintiffs’ position is compelling. Contrary to Texas’s claims, all plaintiffs have First Amendment rights, insofar as they are speaking in the United States. We have never held that because of their lack of citizenship, non-Americans have no rights under the First Amendment when they speak in the United States, and we will not do so here. The district court properly distinguished USAID v. Alliance for Open Society International, Inc., as ap-plying to “foreign organizations operating abroad.” 140 S. Ct. 2082, 2088 (2020). Here, we have foreign organizations speaking in the United States. In fact, because H.B. 1181′s health warnings compel only speech to visitors from Texas, the organizations are effectively only speaking in Texas. Thus plaintiffs, domestic and foreign, would suffer irreparable harm without this injunction against the health warnings. Next, we turn to the balance of harms and public interests. The state’s interests and the public’s have merged here. Texas asserts that the harm to the public interest of not enforcing a state’s law tips the balance of equities in Texas’s favor. Plaintiffs respond by explaining that, should plaintiffs ultimately prevail, Texas’s sovereign immunity will prevent them from recovering compliance and litigation costs. But the government suffers no injury when a court prevents it from enforcing an unlawful law.[75] Thus, the balance of harms and the public interest weigh in plaintiffs’ favor as to the health warnings. * * * * * For the above reasons, we VACATE the stay, VACATE the injunction as to the age-verification requirement, and AFFIRM the injunction as to the health warnings. Patrick E. Higginbotham, Circuit Judge, dissenting in part and concurring in part: Advocate that he was, James Madison penned the Amendments to the work of the constitutional convention, ordering them in his perceived scale of value to their adoption by the States. Its valence in ratication aside, inherent in the First Amendment’s seminal collage of religion, speech, press, assembly, and petition lie the seeds of a signal commitment to individual autonomy, yet to be realized, and in many ways a child itself until the 20th century when the sense of its embrace of individual worth soon became palpable. The years that followed vindicated Madison’s placement of the First Amendment with its rails for the paths of government, married to the individual’s right of identity and self-expression in their myriad forms. At its core, the right of free speech moves with and nds expression in changes of technology, with accompanying eorts by Congress and state legislatures to nd accommodation. In this dynamic mix, Texas has the right—indeed, the obligation—to protect its children. And consistent with this task, it is a given that the State enjoys great latitude in identifying and addressing injury to persons and institutions. Yet implicit in this legal churn remains the core principle that state power must operate within the sinews of the First Amendment, ever a challenge to all of government, a challenge requiring government to attend to its defense, ever faithful to Madison’s gage of the reluctance of the States to relinquish their sovereign interests to the forming of the Union, a concern he further responded to with the assuring language that “Congress shall make no law.”1 I. On August 4, 2023, a group of online pornography websites, performers, and advocates sued the State of Texas to enjoin H.B. 1181 from going into eect on September 1, 2023, contending that the law violated the First Amendment of the United States Constitution and was preempted by § 230 of the Communications Decency Act.2 The district court determined H.B. 1181 was subject to strict scrutiny as a content-based restriction on speech because it applies only to websites that contain “sexual material harmful to minors.” All parties acknowledged the State’s compelling interest in protecting minors, but the court held that H.B. 1181 likely failed strict scrutiny review because the statute was neither narrowly tailored nor the least restrictive means of advancing the State’s interest. The district court concluded that Plaintis demonstrated a substantial likelihood of success on the merits of their claim and granted the preliminary injunction. Texas appealed the district court’s order, arguing the court erred in applying strict scrutiny and nding that Plaintis demonstrated a substantial likelihood of success on the merits. In my view, H.B. 1181 is subject to strict scrutiny, and nding no error in the district court’s factual ndings, I would arm and allow the parties to develop the factual record in the proper forum—trial. II. First we ask whether the district court applied the appropriate level of scrutiny in nding a substantial likelihood of success on the merits, a question of constitutional law reviewed de novo.3 The State argues the district court erred in applying strict scrutiny because H.B. 1181 applies only to speech that is “obscene” for minors undeserving of First Amendment protection and, relying on Ginsberg v. New York, that rational basis is the appropriate metric.4 The majority agrees, concluding that rational basis review applies because the State’s compelling interest in protecting children justies regulating “the distribution to minors of materials obscene for minors.” I disagree. To these eyes, H.B. 1181 cannot be reasonably read to reach only obscene speech in the hands of minors. Although the statute incorporates Miller v. California‘s denition of obscenity, H.B. 1181 limits access to materials that may be denied to minors but remain constitutionally protected speech for adults. It follows that the law must face strict scrutiny review because it limits adults’ access to protected speech using a content-based distinction—whether that speech is harmful to minors. A. “At the heart of the First Amendment lies the principle that each person should decide for himself or herself the ideas and beliefs deserving of expression, consideration, and adherence.”5 It is well established that, except for several narrow categories of speech deemed unworthy of First Amendment protection, all speech is protected by the First Amendment and infringement upon protected speech receives heightened scrutiny. No government can “prescribe what shall be orthodox in politics, nationalism, religion, or other matters of opinion,”6 nor can our government regulate speech “because of disapproval of the ideas expressed.”7 Content-based regulations are then presumed invalid, and the courts must apply heightened scrutiny to laws that disadvantage speech because of its content, those that compel speech, and statutes that infringe upon adults’ constitutionally protected speech.8 The Supreme Court armed these principles in four cases since Ginsberg: Sable Communications of California, Inc. v. F.C.C., Reno v. American Civil Liberties Union, United States v. Playboy Entertainment Group, Inc., and Ashcroft v. American Civil Liberties Union (“Ashcroft II“).9 Each of these cases recognized the government’s compelling interest in protecting children from obscene materials but nevertheless evaluated the laws at issue under strict scrutiny because the law infringed constitutionally protected speech or imposed distinctions based on content.10 Stepping past this precedent, the majority’s new rule unjustifiably places the government’s interest upon a pedestal unsupported by Supreme Court precedent. B. I turn now to the question of whether H.B. 1181 applies only to obscene speech, one of the few categories of speech outside the umbrella of protection aorded by the First Amendment.11 Relevant here, the law applies only to “commercial entit[ies] that knowingly and intentionally publish[] or distribute[] material on an Internet website, including a social media platform, more than one-third of which is sexual material harmful to minors.”12 “Sexual material harmful to minors” is dened as “any material” that: (A) the average person applying contemporary community standards would nd, taking the material as a whole and with respect to minors, is designed to appeal to or pander to the prurient interest; (B) in a manner patently oensive with respect to minors, exploits, is devoted to, or principally consists of descriptions of actual, simulated, or animated displays or depictions of: (i) a person’s pubic hair, anus, or genitals or the nipple of the female breast; (ii) touching, caressing, or fondling of nipples, breasts, buttocks, anuses, or genitals; or (iii) sexual intercourse, masturbation, sodomy, bestiality, oral copulation, agellation, excretory functions, exhibitions, or any other sexual act; and (C) taken as a whole, lacks serious literary, artistic, political, or scientic value for minors.13 Although obscene speech lies outside the First Amendment’s umbrella of protection, not all sexual expression is obscene.14 Indeed, “sexual expression which is indecent but not obscene is protected by the First Amendment.”15 What Plaintis refer to as “exclusively ‘soft core’ nude modeling,” for example, constitutes non-obscene sexual expression, as would many romance novels, or—to use another example from the brieng— Marlon Brando movies. And protected sexual expression encompasses materials that are appropriate for adults but inappropriate for minors. For example, scenes from the popular show “Game of Thrones,” the 1985 lm “The Color Purple,” or the 2011 lm “The Girl with the Dragon Tattoo” all contain “depictions” of sexual intercourse that may be “patently oensive” to young minors and regulated under H.B. 1181, but still oer artistic or cinematic value for adults. While I agree with the majority that H.B. 1181′s plain text applies only to “sexual material harmful to minors,”16 the statute cannot be reasonably read to regulate only obscene content. In the words of the district court, H.B. 1181 goes “beyond obscene materials” and “regulates all content that is prurient, oensive, and without value to minors.”17 In doing so, the law infringes upon adults’ protected sexually expressive speech. Ultimately, the text does not support the argument that H.B. 1181 regulates only obscene speech.18 H.B. 1181 regulates all material harmful to minors, which necessarily encompasses non-obscene, sexually expressive— and constitutionally protected—speech for adults. Thus, H.B. 1181 limits access to constitutionally protected speech, regardless of whether the viewer is a minor. Such action “is to burn the house to roast the pig.”19 C. As the regulated speech falls under the First Amendment’s umbrella of protection, the issue is whether the district court properly found H.B. 1181 subject to strict scrutiny, a question of law reviewed de novo.20 The majority nds H.B. 1181 subject only to rational basis review. I disagree. 1. Content-based restrictions on protected speech are presumptively unconstitutional, valid only if the government proves they are narrowly tailored to further a compelling interest.21 By the statute’s plain language, H.B. 1181 applies only to websites with content “more than one-third of which is sexual material harmful to minors.“22 Because H.B. 1181 regulates only a particular type of speech, “[t]he speech in question is dened by its content; and the statute which seeks to restrict it is content based.”23 As such, H.B. 1181 is subject to strict scrutiny. 2. The district court found the State “largely concede[d]” that strict scrutiny should apply, but looking to Ginsberg, the State now asks this Court to nd that this content-based restriction does not warrant strict scrutiny.24 While the majority credits this argument, I cannot—for Ginsberg does not here call for rational basis review, and the Supreme Court has unswervingly applied strict scrutiny to content-based regulations that limit adults’ access to protected speech. In Ginsberg, the Supreme Court upheld a New York criminal obscenity statute prohibiting the knowing sale of obscene materials to minors.25 Ginsberg was convicted of violating the statute after he sold two “girlie magazines” to a sixteen-year-old. Ginsberg asserted that the New York statute violated the First Amendment because “the constitutional freedom of expression secured to a citizen to read or see material concerned with sex cannot be made to depend upon whether the citizen is an adult or a minor.”26 He went on to argue “that the denial to minors under 17 of access to material condemned by [the statute], insofar as that material is not obscene for persons 17 years of age or older, constitutes an unconstitutional deprivation of protected liberty,” which Ginsberg likened to the deprivations of liberty recognized in Meyer v. State of Nebraska, Pierce v. Society of Sisters, and West Virginia State Board of Education v. Barnette.27 The Supreme Court disagreed, focusing on the fact that the prosecution concerned a single sale in Ginsberg’s store to a minor. Despite observing that the magazines were “not obscene for adults,” the Court held the New York regulation did not invade the “minors’ constitutionally protected freedoms.”28 Explaining that “the power of the state to control the conduct of children reaches beyond the scope of its authority over adults,” the Court found the law rationally related to the State’s interest in protecting minors, and upheld Ginsberg’s conviction.29 Ginsberg‘s force here is its recognition of a state’s power to regulate minors in ways it could not regulate adults. But this overriding power to protect children does not answer our essential question: whether H.B. 1181 imposes a content-based restriction or causes an “unnecessarily broad suppression of speech addressed to adults.”30 If so, “the answer should be clear: The standard is strict scrutiny.”31 Ginsberg is further inapplicable because application of rational basis rested on the notion that minors have more limited First Amendment rights than adults32 and because the statute did not infringe upon adults’ access to the same materials.33 Notably, Plaintis here, unlike Ginsberg, do not seek greater access for minors to these materials. And the New York statute at issue in Ginsberg did not burden the free speech interests of adults, but H.B. 1181 does; H.B. 1181 requires that adults comply with the age verication procedure and view the required health disclosures before accessing protected speech. Therefore, Ginsberg‘s justication for rational basis review—that minors have more limited First Amendment rights than adults—has no purchase here, as we are dealing with a challenge to an adult’s ability to access constitutionally protected materials on the ubiquitous internet, not over-the-counter magazine sales in a drug store. The majority cites to Erznoznik v. City of Jacksonville for the broad contention that more stringent regulations of content are permissible for minors without any regard to burdens placed on adult speech. Erznoznik invalidated a Jacksonville ordinance that prohibited drive-in theaters from exhibiting films showing bare buttocks or breasts.34 In the majority’s view, had the definition of obscenity been narrowed to only depictions obscene for children, the Erznoznik Court would have taken no issue with the burdens imposed on adults. With respect, the Court did consider the burdens on adults. Specifically, the Court noted that “the deterrent effect of this ordinance [was] both real and substantial,” as it “applie[d] to all persons employed by or connected with drive-in theaters” and forced “the owners and operators of these theaters” to either “restrict their movie offerings or construct adequate protective fencing which may be extremely expensive of even physically impracticable” in order to avoid prosecution.35 It also explained that the ordinance deterred theaters “from showing movies containing any nudity, however innocent or even educational.” Contrary to the majority’s view, and as we have aways done, Erznoznik requires an exacting comparison of the state interest involved (including the interest in protecting children) against the methods chosen to eectuate the state’s interest and its eects on the broader adult population.36 It is no failure of advocacy that the State has cited to no case since Ginsberg in which the Supreme Court applied rational basis review to regulations impinging adults’ access to protected speech.37 No such case exists. Instead, since Ginsberg, the Supreme Court has consistently applied strict scrutiny to content-based regulations that infringe upon adults’ protected speech. In Sable, the Supreme Court addressed § 223(b) of the Communications Decency Act of 1934, as amended in 1988.38 Section 223 imposed a blanket prohibition on indecent and obscene interstate commercial telephone messages, including prerecorded “dial-a-porn” services.39 The CDA’s proscriptions were justied based on the government’s interest in protecting children from harmful materials.40 The Supreme Court upheld § 223(b)’s ban of obscene commercial telephone messages but struck the limitations on “indecent” communications as stepping on adults’ access to constitutionally protected sexual expression.41 Citing Ginsberg, the Court acknowledged the federal government’s compelling interest in protecting the “physical and psychological well-being of minors” but nevertheless required the statute be “narrowly drawn.”42 In short, the Court applied strict scrutiny because “the statute’s denial of adult access to such messages far exceeds that which is necessary to serve the compelling interest of preventing minors from being exposed to the messages.”43 The Communications Decency Act drew challenge again in Reno v. American Civil Liberties Union.44 In Reno, the Court engaged § 223 (a) and (d), which criminalized the “indecent transmission” and “patently oensive display” of “obscene or indecent” messages to any recipient under 18 years of age.45 As the Supreme Court had not yet articulated a standard for restrictions on internet communications, the rst question for the Court was the appropriate level of scrutiny. The government urged the Court to apply rational basis review like Ginsberg, or alternatively, intermediate scrutiny as the Court had done in F.C.C. v. Pacica Foundation and City of Renton v. Playtime Theatres, Inc.46 The Supreme Court disagreed with the government. First, it found that the CDA had four important textual dierences from the New York statute at issue in Ginsberg.47 Then, the Reno Court held Pacica inapplicable because the Pacica broadcast regulation was narrower than the CDA, did not involve a punitive component, and concerned a medium which, as a historical matter, had “received the most limited First Amendment protection.”48 Finally, Reno found that the “time, place, and manner” restrictions in Playtime Theatres were inapposite because the CDA was designed to target the “primary eects” of oensive speech and not the “secondary eects” at issue in Playtime Theatres.49 The Court concluded that “these precedents, then, surely do not require us to uphold the CDA and are fully consistent with the application of the most stringent review of its provisions.”50 The Court went on to distinguish “cyberspace” from traditional mediums of expression.51 In particular, the Court found “the vast democratic forums of the Internet” had never been subject to the same degree of “government supervision and regulation that has attended the broadcast industry.”52 Moreover, the internet was not as “invasive” as radio, in that internet communications “do not ‘invade’ an individual’s home or appear on one’s computer screen unbidden.”53 Indeed, because internet users must take “armative steps” to access sexually explicit content, the Court noted that “users seldom encounter content by accident” and the “odds are slim that a user would come across a sexually explicit sight by accident.”54 Ultimately, the Court in Reno found that “our cases provide no basis for qualifying the level of First Amendment scrutiny that should be applied to this medium.”55 As a content-based regulation of speech, the CDA faced the “most stringent review of its provisions” and failed, as it was not narrowly tailored and less restrictive alternatives were available.56 Displeased with the Reno decision, Congress enacted the Child Online Protection Act (“COPA”).57 The COPA imposed a $50,000 ne or up to six months’ imprisonment for anyone who knowingly used the internet to make “any communication for commercial purposes that is available to any minor and that includes any material that is harmful to minors.”58 The statute distinguished permissible from impermissible communications based on whether they were “harmful to minors.”59 The COPA dened “harmful to minors” by adopting Miller‘s denition of obscenity.60 In Ashcroft II, several internet service providers, “Web speakers[,] and others concerned with protecting the freedom of speech” challenged the COPA as a content-based speech restriction.61 As in Sable and Reno, the government asserted the COPA was constitutional given the government’s interest in protecting minors from harmful materials.62 The district court applied strict scrutiny, found the COPA was likely to deter adults from accessing protected speech, and concluded that the government could not “meet its burden to prove that COPA is the least restrictive means available to achieve the goal of restricting the access of minors to harmful material.”63 Accordingly, the district court preliminarily enjoined the statute’s enforcement.64 The Supreme Court agreed and armed the district court.65 In doing so, Ashcroft II restated Reno‘s rule that strict scrutiny applied to statutes that “eectively suppress[] a large amount of speech that adults have a constitutional right to receive and to address to one another.”66 Ashcroft II reiterated that “content-based restrictions on speech [are] presumed invalid, and that the Government bear[s] the burden of showing their constitutionality.”67 Ultimately, the Court upheld the preliminary injunction after nding a number of plausible, less restrictive alternatives, including blocking and ltering software.68 The Supreme Court has consistently applied strict scrutiny to content-based restrictions that impair adults’ access to protected speech. Contrary to the majority’s views, Ginsberg is distinguishable and does not change this analysis. H.B. 1181 imposes a content-based restriction on speech that burdens adults’ access to that speech. Bound by Sable, Reno, and Ashcroft II, this Court must apply strict scrutiny. 3. In an eort to escape the Supreme Court’s insistence upon strict scrutiny for content-based restrictions, the majority—and the State—would dierentiate H.B. 1181 from the CDA in Reno and the COPA in Ashcroft II. These distinctions are unpersuasive. i. First, the State attempts to distinguish this case from Reno, arguing that H.B. 1181 is narrower than the CDA and more closely tracks Miller‘s denition of obscenity.69 Indeed, Reno found the CDA’s denition of obscenity to be impermissibly broader than Miller‘s for several reasons: Miller‘s denition of obscenity was limited to conduct dened by state law, while the CDA’s was not; and Miller‘s denition was limited to sexual conduct, whereas the CDA also covered excretory activities and “organs of both a sexual and excretory nature.”70 H.B. 1181is strikingly similar to the CDA and, in some ways, goes even further. Like the CDA, H.B. 1181 does not limit regulated speech to conduct proscribed by Texas law.71 Like the CDA, H.B. 1181 regulates more than just “sexual conduct.”72 The CDA prohibited speech regarding “excretory activities” as well as “organs” of both a sexual and excretory nature,73 and H.B. 1181 similarly restricts depictions of “pubic hair” and “the nipple of the female breast.”74 By its text, H.B. 1181 goes further than the CDA regarding the format of depictions it covers, as it applies to “descriptions of actual, simulated, or animated displays or depictions” of specied body parts, conduct, and undened “exhibitions,”75 while the CDA applied, inter alia, to “image[s].”76 In essence, Texas’s contention that H.B. 1181 closely tracks Miller fails to persuade. Lastly, the majority discounts Reno on the basis that it did not distinguish the internet from in-person communications, to which I say: read Reno. The opinion reviewed the levels of scrutiny applied across numerous cases and mediums—including those applied in Ginsberg, Pacica, Playtime Theatres, and Sable—before concluding “our cases provide no basis for qualifying the level of First Amendment scrutiny that should be applied to this medium [the internet].”77 ii. Texas’s attempt to distinguish Ashcroft II is similarly awed. As an initial matter, the State understates the similarities between H.B. 1181 and the COPA. Where the COPA distinguished content “harmful to minors,” H.B. 1181 uses the phrase “sexual material harmful to minors.” The statutory denitions of “harmful to minors” and “sexual material harmful to minors” are, however, functionally identical.78 These similarities are not supercial, they are substantive. Therefore, as in Ashcroft II, strict scrutiny applies. Moreover, Texas’s argument that Brown v. Entertainment Merchants Association rearmed Ginsberg, while condoned by the majority, stretches Brown‘s holding.79 Texas is correct that Brown armed that a state has a compelling interest in protecting minors from obscenity.80 But that is where the similarities to Ginsberg end. The Brown Court did not apply Ginsberg‘s rational basis review. Instead, the Court found the California statute at issue constituted a content-based restriction of protected speech and applied strict scrutiny.81 Finally, the majority’s critique that Ashcroft II contained “startling omissions” regarding its analytical framework ignores that the Supreme Court itself previously found that strict scrutiny applied.82 In Ashcroft II, the Supreme Court simply treated it as a self-evident proposition that strict scrutiny applied. This Court cannot fault Ashcroft II for applying the level of scrutiny clearly established by Sable and Reno, or for declining to engage in repetitive analysis. And the majority’s implication that the Supreme Court knowingly applied the wrong level of scrutiny merely because the issue was not “jurisdictional” needs no response. * * * The majority’s attempts to distinguish Ginsberg from Sable, Reno, and Ashcroft II are unconvincing. H.B. 1181 implements a content-based distinction: it applies only if over one third of an entity’s commercial content is “harmful to minors.” While the majority claims “the statute in Sable swept in a much larger swath of speech than the speech targeted here,” it fails to explain how § 224(b)’s ban on “indecent” material is broader than that regulated by H.B. 1181. Further, the Supreme Court has rejected the majority’s contention that there is a dierence between regulatory burdens and bans: “It is of no moment that the statute does not impose a complete prohibition. The distinction between laws burdening and laws banning speech is but a matter of degree.”83 In sum, I agree with the majority that Ginsberg remains good law and indubitably recognizes the government’s power to protect children from age-inappropriate materials. But content-based laws that infringe upon protected speech receive strict scrutiny. H.B. 1181 imposes a content-based restriction on speech that burdens adults’ access to that speech and is subject to strict scrutiny. III. To prevent minors from accessing potentially harmful speech, H.B. 1181 “suppresses a large amount of speech that adults have a constitutional right to receive and to address to one another.”84 Under strict scrutiny, the statute must be narrowly drawn to further Texas’s compelling interest and the restriction must amount to the “least restrictive means to further the articulated interest.”85 There was no challenge before the district court to the State’s compelling interest in protecting minors from inappropriate, explicit content. Rather, the district court held that H.B. 1181 failed strict scrutiny review because it was neither narrowly tailored nor the least restrictive means of achieving the State’s interest. Hence, the district court concluded Plaintis demonstrated a substantial likelihood of success on the merits of their claim and granted the preliminary injunction. We review these ndings for clear error.86 A. First, the age verication component of H.B. 1181 requires regulated commercial entities to implement “reasonable age verification methods” to ensure that all website users are over the age of 18.87 “Reasonable age verication methods” include requiring website viewers to either “provide digital identication” or “comply with a commercial age verication system that veries age” using either government-issued identication or “a commercially reasonable method that relies on public or private transactional data to verify the age of an individual.”88 The district court found that this provision of the statute was not narrowly tailored, as it was underinclusive, vague, and overbroad in its regulation of adults’ protected speech. I agree. 1. i. The district court found H.B. 1181′s age verication mandate “severely underinclusive” because the law does not regulate sites that are “most likely to serve as a gateway to pornography use.” This conclusion rested on expert declarations that highlighted the types of entities left unregulated by H.B. 1181. These reports emphasized that internet service providers are explicitly exempt from the statute.89 Consequently, this exemption “ignores visual search[es], much of which is sexually explicit or pornographic,” and could be easily accessed by children after a simple “misspelled search.” Further relying on the experts’ reports, the district court concluded that social media platforms are also implicitly exempted from the statute because “they likely do not distribute at least one-third sexual material.” The district court found this exemption problematic because some social media sites, like Reddit or Tumblr, contain “entire communities and forums . . . dedicated to posting online pornography with no regulation under H.B. 1181.” Likewise, the district court found that social media websites like Instagram and Facebook “can show material which is sexually explicit for minors without compelled age verication.” Ultimately, the district court concluded that, because of these exceptions, H.B. 1181 “fails to reduce the online pornography that is most readily available to minors.” At this juncture, Texas has failed to persuade that this nding was clearly erroneous. The State argues the age verication requirement is not underinclusive but, instead, only targets sites “whose business model is signicantly driven by distributing sexual material harmful to minors,” unlike search engines and social media websites that have either taken steps to protect minors or “do not seek prot from peddling depictions of bestiality and sexual assault.” This argument fails at the starting gate because it was not made before the district court and is thus forfeited on appeal. Equally important, the State oered no evidence that the Texas legislature tailored H.B. 1181 to address pornography websites because their “business model” is tailored to selling pornography to minors. Without evidence in the record, this argument cannot stand. Rather, the present record supports the district court’s conclusion that while H.B. 1181 will regulate adult video companies, it “will do little else to prevent children from accessing pornography.” The district court evaluated the provided expert reports and based its conclusions on the evidence before it. I am not left with a “definite and firm conviction that a mistake has been committed,”90 a conclusion reinforced by the “great deference” owed to “the findings of the trial court with respect to duly admitted expert testimony.”91 ii. Turning to vagueness, the district court further held that the age verication component of H.B. 1181 was “problematic because of several key ambiguities” in the statutory language. To begin, the court evaluated § 129B.002and found that by lumping together young minors with those near the age of majority, the statute overlooked the reality that material “harmful to a younger minor is vastly dierent . . . than material that is harmful to a minor just shy of” majority.92 Similar problems stemmed from the statute’s failure to dene “serious literary, artistic, political, or scientic value for minors.”93 As such value varies with age group, the district court found that H.B. 1181 would chill constitutionally protected speech. Additionally, the district court found H.B. 1181′s scope was “subject to multiple interpretations as to the scope of its liability” because it was unclear whether the “one-third” requirement modied “material” or “website.” Finally, the court was concerned that H.B. 1181 did not dene a “commercially reasonable method” of age verication. The district court indicated these vague provisions were problematic and spoke “to the statute’s broad tailoring.”94 In response to these challenges, the State asserts that “one-third” modies “website” and argues any problems with vagueness could be resolved through “certication.”95 Even if we accepted this interpretation, the State does not address the district court’s remaining concerns regarding the denitions of a “commercially reasonable method” of age verication or “serious literary, artistic, political, or scientic value for minors.” More to the point, the district court did not rest its holding on a vagueness challenge. The purchase of its observations was that H.B. 1181 was not narrowly tailored. iii. In granting the preliminary injunction, the district court found thatthe age verication requirements were impermissibly broad and would infringe upon adults’ constitutionally protected speech. The court determined that the statute was “largely identical” to the COPA, found unconstitutional in Ashcroft II; likely to chill substantial speech for adults, as it does not require the government to delete user data and “risk[ed] forcing individuals to divulge specic details of their sexuality to the state government to gain access to certain speech”; and eectively “reduce[d] the adult population to only what is t for children.”96 The State disputes these conclusions and argues that the statute “does not prohibit any speech, only [requires] that [Plaintis] check the ages of their customers,” and will not chill speech because the age verication requirements “preserve[] online anonymity.” These arguments fail for four reasons. First, that H.B. 1181 “does not prohibit any speech” but merely restricts access to speech is an empty argument because “[t]he distinction between laws burdening and laws banning speech is but a matter of degree.”97 Second, H.B. 1181 encompasses nearly all salacious material because it may be harmful to young minors, see supra Section II.B, and as such, the statute restricts adults’ access to both obscene and non-obscene speech. Third, the State does not address the district court’s concerns that government entities and third-party intermediaries are not required to delete users’ data. H.B. 1181 prohibits commercial entities and third parties performing age verication from retaining identifying information, but the bill imposes no burden on governmental entities nor “any intermediary between the commercial websites and the third-party veriers” to do the same.98 Simply claiming that the “age verication preserves online anonymity” does not make it so. Finally, this response ignores the “special First Amendment concerns” of the chilling eects on speech when the state government can log and track adults’ access to sexual material.99 It is canon that overbroad regulations “have the potential to chill, or deter, speech outside their boundaries.”100 The risk of censorship is particularly high for regulations of obscenity and sexual expression, which are “often separated . . . only by a dim and uncertain line.”101 Here, the record supports the district court’s conclusion that the age verication mandate will chill protected speech; moreover, it suggests the majority’s claim that H.B. 1181 has a “de minimis eect on privacy” understates the privacy issues at play. The mandate requires adults to armatively identify themselves by providing government identication before accessing desired content. This requirement “deters adults’ access to legal sexually explicit material” and goes “far beyond the interest of protecting minors.” Because neither the government nor “any intermediary” is required to delete information obtained, the district court found that adults would be “particularly concerned about accessing controversial speech when the state government can log and track that access.” Texas provides no persuasive rebuttal. 2. A content-based statute that impermissibly burdens adults’ protected speech “is unacceptable if less restrictive alternatives would be at least as eective in achieving the legitimate purpose that the statute was enacted to serve.”102 The State bears the burden of proving its chosen method is constitutional,103 which often requires proof that the legislature considered alternative, less restrictive means.104 The State has oered no evidence on this latter point and failed to meet its burden. Before the district court, Plaintis oered two less restrictive methods to shield children from inappropriate sexual content: (1) requiring internet service providers, or ISPs, to block specied content until adults opt-out of the block; and (2) “content ltering” by implementing adult controls on children’s devices. The district court found that Plaintis’ proposed alternatives were less restrictive of adult speech, citing the State’s own exhibits that indicated “content ltering” was the “most eective method of limiting any harm to minors.” The district court concluded that content blocks would address the “under-inclusivity issue” and “comport[] with the notion that parents, not the government, should make key decisions on how to raise their children.” On appeal, Texas contends that content blocking and ltering are not eective, less restrictive alternatives to age verication. It argues that the proposed age verication methods do not require the disclosure of sensitive information and that companies could use techniques such as “sele matching” or age estimation.105 This argument eectively asks the Court to reweigh the evidence and review the district court’s ndings de novo. We may not do so. The State made these arguments before the district court, which, in its opinion, discussed the State’s expert report and explained why it found the contents unpersuasive. The district court considered and rejected these arguments, and I would nd no clear error in its determination that the Texas legislature failed to consider less restrictive means to accomplish its goals. B. Next, I turn to H.B. 1181′s health warnings provision. H.B. 1181 requires regulated commercial entities to post three warnings purportedly authored by Texas Health and Human Services, as well as a phone number for the U.S. Substance Abuse Mental Health Services Administration.106 I would arm the district court’s enjoining of the health warnings for the reasons articulated by the district court. Although I would apply strict scrutiny, I concur with the majority’s judgment that the compelled disclosures do not survive scrutiny. At this junction, the notice requirements are unenforceable. IV. The district court held that the Plaintis who merely host third-party content—MG Freesites LTD, WebGroup Czech Republic, NKL Associates, s.r.o., and Media Me SRL—are entitled to an injunction under Section 230 of the CDA. Specically, the district court explained that Section 230 immunizes these Plaintis against H.B. 1181′s enforcement because the statute purports to penalize them for hosting sexual content created by others.107 This analysis is sound, faithful to the statutory text of the CDA and binding case law. Relevant here, the CDA states that “[n]o provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider.”108 The law further provides that “no cause of action may be brought and no liability may be imposed under any State or local law that is inconsistent with this section.”109 The CDA is intended “to promote rather than chill Internet speech. . . . [paving] the way for a robust new forum for public speech as well as a trillion-dollar industry centered around user-generated content.”110 The State rst asserts Section 230 does not preempt H.B. 1181 because the statute “neither holds Pornographers liable for third-party content . . . nor imposes liability for good-faith measures to restrict access to oensive material.” Second, the State argues Plaintis cannot take advantage of the immunity provision because they are “responsible, in whole or in part, for the creation or development of information provided through the Internet.” Third, the State contends foreign corporations may not assert First Amendment claims because they do not have constitutional rights. Lastly, the State argues that to the extent Section 230 protects Plaintis, they cannot assert First Amendment claims. The State’s rst two arguments are foreclosed be Doe v. MySpace, wherein this Court noted that “Congress provided broad immunity under the CDA to Web-based service providers for all claims stemming from their publication of information created by third parties[.]“111 Although “[p]arties complaining that they were harmed by a Web site’s [sic] publication of user-generated content . . . may sue the third-party user who generated the content,” they may not sue “the interactive computer service that enabled them to publish the content online.”112 In Doe, the plainti sued MySpace for negligence in allegedly failing to take precautions to prevent sexual predators from communicating with minors and specically complained that MySpace failed to implement age-verication software.112 113But this Court held that the CDA barred the claim: “notwithstanding [plainti's] assertion that they only seek to hold MySpace liable for its failure to implement measures that would have prevented” the plainti from communicating with the sexual predator, the “allegations are merely another way of claiming that MySpace was liable for publishing the communications.”114 The State tries to distinguish this case from Doe, but its arguments are meritless. First, it asserts that “H.B. 1181 seeks to hold pornographic sites responsible for failing to protect minors from content they generate and promote, not just third-party content.” However, the district court considered thoroughly the various Plaintis’ business models and found that some of the Plaintis, including “WebGroup, which operates XVideos, only host[] third-party content, and therefore [are] entitled to Section 230 protection.” These ndings of fact are to be upheld unless they are clearly erroneous. I nd no error. Second, the State argues that after Plaintis “use the age-verication methods . . . their job is done,” implying that H.B. 1181 does not directly penalize parties for hosting sexual content. But this Court held explicitly in Doe that requiring websites that only host third-party content to implement age-verication measures violates Section 230. The CDA immunizes platforms from all liability associated with hosting third-party content and it preempts all statutes inconsistent with this mandate. H.B. 1181 imposes severe civil liability, mandatory disclosures, and age verication requirements based on the presence of third-party content. That websites will be safe from H.B. 1181′s signicant civil penalties if they implement the required age-verication system is no answer. Third, the State argues Plaintis are still “responsible” for the content they host such that they lose Section 230 protection. Specically, the State asserts that Plaintis are not entitled to immunity under Section 230 because they “promote content in special parts of their websites” and “armatively license and advertise rather than passively host content.” But this argument is also unpersuasive. That licensing and promoting content created by third parties causes developers to lose Section 230 immunity is not supported by any precedent of this Court. It would also swallow the CDA’s rule of immunity and undermine Congress’s clear goals to insulate website owners and developers from liability stemming from the publication of third-party content. Next, the State’s argument that foreign corporations do not have constitutional rights, such that the foreign Plaintis cannot benet from First Amendment protections, fails to persuade. The State bases its argument on Agency for International Development v. Alliance for Open Society International (“AOSI II“), which held that foreign organizations operating outside the United States are not entitled to the protections of the United States Constitution.115 But AOSI II has no bite in this case where “the recipients of [Plaintis'] speech and speech-related conduct” are in the United States.116 Instead, “[a]s the [Supreme] Court has recognized, foreign citizens in the United States may enjoy certain constitutional rights.”117 Moreover, this argument overlooks Plaintis’ abilities to vindicate the First Amendment rights of their U.S. site visitors.118 Lastly, I cannot agree with the State’s argument that those parties asserting Section 230 immunity cannot also bring First Amendment claims. The State relies on this Court’s language in NetChoice to support its contention: “§ 230 reects Congress’s judgment that the [websites] are not acting as speakers or publishers when they host user-submitted content.”119 The State argues Plaintis’ attempt to assert a Section 230 claim on behalf of themselves and First Amendment claims on behalf of their customers fails because “the challenge to HB [sic] 1181′s notice requirement is not asserted on behalf of customers [] and the age-verication claim is not limited to the First Amendment rights of their customers.” As an initial matter, the district court never discussed whether the health notices requirement was barred by Section 230. So, I will not address that question. With regard to the age-verication requirement, I would nd no error in the district court’s analysis. First, the district court found that Plainti Free Speech Coalition has associational standing; thus, Plaintis may bring a First Amendment facial challenge to H.B. 1181 and argue that it is an overbroad content-based regulation.120 Second, although Plaintis entitled to Section 230 immunity are “carriers” and not “creators” of speech, they may still assert First Amendment claims on behalf of adults wishing to view protected sexual content. As the district court explained: Beyond their own First Amendment injuries, Plaintis have standing for their overbreadth challenge. Broadrick v. Oklahoma, 413 U.S. 601, 612 (1973) (“Litigants, therefore, are permitted to challenge a statute not because their own rights of free expression are violated, but because of a judicial prediction or assumption that the statute’s very existence may cause others not before the court to refrain from constitutionally protected speech or expression.”); Sec. of State of Md. v. Joseph H. Munson Co., Inc., 467 U.S. 947, 956 (1984) (“[W]hen there is a danger of chilling free speech, the concern that constitutional adjudication be avoided whenever possible may be outweighed by society’s interest in having the statute challenged.”). The district court did not err when it held those Plaintis that only hosted third-party content were entitled to Section 230 immunity, nor did it err when it held that the First Amendment claims regarding H.B. 1181′s age verication requirement may exist alongside Section 230 liability claims. Both rulings comport with the text of Section 230 and Doe v. MySpace. V. It is significant to note that H.B. 1181 fails exacting scrutiny at this stage in large part for want of evidence. Aside from a single “Bill Analysis” completed by the Texas Senate Research Center—which the State did not provide to the district court but can nonetheless be considered—the record is bereft of evidence responsive to the burdens of strict scrutiny.121 That is not to say that the legislature did not consider alternatives or that the State will be unable to provide this and other evidence at trial. At this junction and with this record, however, I would not upset the district court’s finding that Plaintiffs have a substantial likelihood of success on the merits. That H.B. 1181 now fails strict scrutiny does not foreclose further attempts by the State to legislate on this issue. To the contrary, Texas has the right—and to these eyes, the obligation—to protect its minors, and in doing so, it must have the means to frustrate their access to pornographic materials consistent with the First Amendment and Section 230. The State may decide that the first step of this march is to place the onus on parents to monitor their children’s viewing habits and provide them the tools to do so. Indeed, to quote the State: “parents are the first line of defense.” Implicit in this reality is that educating parents addresses the State’s concern that children “are often more adept at circumventing such software than [parents] are at issuing it”122 without infringing adults’ access to protected speech. Continuing on that path, the State can look again to requiring electronic filters on devices, as considered in the “Bill Analysis” conducted by the Senate,123 such as blocking software or shared-data applications that provide parents with affordable access to their children’s devices. While the decision to travel a particular legislative pathway is beyond this Court’s compass, illuminating both the available legal options and their outer rails is not. At the end of the day, the goals of H.B. 1181 will not be thwarted. Rather, legal pathways for its eventual success will be laid. For now, I see the pathways to statutory protection inevitably turning the focus to parents, arming them with means of protecting their children. The interest of the sellers of pornography and the interest of the State here converge—both seek the benefit of technology that will enable websites to bypass the eyes of children in their passage of material to adult purchasers. The sellers of the product have powerful incentives to find the means, and we have the proper forum for their display—a trial. Whether such technology exists and its utility are questions of fact for trial. I note only that the websites’ ability—or lack thereof—to provide such a product would be telling. As I am persuaded that, on the record before him, the able veteran district judge did not err in finding a likelihood of success on the merits, I would affirm the grant of a preliminary injunction and return the case to the district court for further proceedings. Facts matter. Facts decide cases and trial is their proper forum.