NetChoice v. Tim Griffin, in his Official Capacity as Attorney General of Arkansas et al.

• Court: District Court, W.D. Arkansas
• Decided: December 15, 2025
• Docket: 5:25-cv-05140
• Status: Unknown

Opinion

IN THE UNITED STATES DISTRICT COURT WESTERN DISTRICT OF ARKANSAS FAYETTEVILLE DIVISION

NETCHOICE PLAINTIFF

V. CASE NO. 5:25-CV-5140

TIM GRIFFIN, in his Official Capacity as Attorney General of Arkansas et al. DEFENDANTS

MEMORANDUM OPINION AND ORDER TABLE OF CONTENTS

I. BACKGROUND .......................................................................................................... 1 II. LEGAL STANDARD .................................................................................................. 7 III. DISCUSSION ............................................................................................................ 8 A. Severability ........................................................................................................... 8 B. Likelihood of Success on the Merits .................................................................. 9 1. First Amendment ................................................................................................. 9 2. Void for Vagueness ........................................................................................... 25 3. Section 230 Preemption .................................................................................... 29 C. Irreparable Harm ................................................................................................ 31 D. Balance of the Equities and the Public Interest .............................................. 32 IV. CONCLUSION ........................................................................................................ 32 NetChoice, an Internet trade association, asks the Court to preliminarily enjoin enforcement of Arkansas Act 901 of 2025, which imposes liability on social media platforms for certain harmful effects a platform may have on a user. NetChoice argues that Act 901 violates the First Amendment, is void for vagueness, and is preempted by

§ 230 of the Communications Decency Act. For the reasons that follow, NetChoice’s Motion for Preliminary Injunction (Doc. 23) is GRANTED.1 I. BACKGROUND Act 901 has two operative provisions. First, § 1502 prohibits a social media platform from using “a design, algorithm, or feature that the social media platform knows, or should have known through the exercise of reasonable care, causes a user to: (1) Purchase a controlled substance; (2) Develop an eating disorder; (3) Commit or attempt to commit suicide; or (4) Develop or sustain an addiction to the social media platform.”

Id. (codified at Ark. Code Ann. § 4-88-1502). Under the general provisions of Chapter 88, the Attorney General has authority to file civil suits to enforce this provision, and the prosecuting attorneys may charge knowing and willful violations as Class A misdemeanors. Ark. Code Ann. §§ 4-88-103 and -104. Second, § 1503 creates a private right of action entitling a parent to damages and other relief if their child commits or attempts suicide “following exposure to online content

1 The Court held a hearing on the Motion on October 28, 2025. The Court has considered NetChoice’s Brief in Support and Exhibits (Docs. 24–28), Defendants’ Response in Opposition and Exhibits (Docs. 36–37), NetChoice’s Reply (Doc. 40), and the parties’ post-hearing supplements (Docs. 44 & 46). promoting, or otherwise advancing, self-harm or suicide” on a “social media platform that hosted, promoted, shared, or otherwise facilitated the immediate connection between the victim and the content.” Id. (codified at § 4-88-1503(b)). A platform that “knowingly and willfully violates” this provision is also liable for a “civil penalty not to exceed ten thousand

dollars ($10,000) per violation.” Id. § 1503(a). Act 901 defines a “social media platform” as “a business entity or organization that operates an online platform, application, or service that: (A) Is designed to facilitate user-to-user, user-to-group, or user-to-public interaction, expression, or communication; (B) Assigns, utilizes, or relies on a unique identifier, username, profile name, or image that is associated with a specific user account; (C) Provides mechanisms for a user to create an online profile comprised of personally identifiable information or professional information, including without limitation a user's name, username, address, date of birth, educational pedigree, professional details, interests, activities, or connections; (D) Employs features that allow a user to connect, follow, or establish a relationship with other users and creates a network of interactions either in real time or asynchronously, including without limitation virtual likes and dislikes; (E) Generates revenue primarily through user engagement, including without limitation through advertising, user data monetization, or premium content; and (F) Is accessed by Arkansas users.” Id. (codified at § 1501(a)(5)). Which platforms are covered by this definition is not disputed. Many of NetChoice’s members are covered, including many of the usual suspects—Meta (parent of Facebook and Instagram), YouTube, Snap Inc. (parent of Snapchat), Reddit, Pinterest, Nextdoor, and X. See Doc. 26, ¶ 4. There is no doubt that users engage in constitutionally protected speech on these platforms. Packingham v. North Carolina, 582 U.S. 98, 105 (2017) “[S]ocial media users employ these websites to engage in a wide array of protected First Amendment activity . . . .”). Platforms themselves also engage in a range of protected expression on their platforms, including the exercise of “editorial judgment[ ]” to “include

and exclude, organize and prioritize” user-generated content. Moody v. NetChoice, LLC, 603 U.S. 707, 716–18 (2024). Social media platforms as commonly understood and as defined by Act 901 generate revenue through advertising. Platforms are therefore designed to maximize the amount of time users spend on them in order to maximize ad revenue. “[T]he most effective way of maintaining a behavior” is intermittent variable rewards. Vikram R. Bhargava & Manuel Velasquez, Ethics of the Attention Economy: The Problem of Social Media Addiction, 31 Bus. Ethics Q. 321, 327 (2021) (Doc. 37-15). Platforms therefore make use of intermittent variable rewards to encourage users to spend more time online. “Intermittent” means the number of times a user must engage in a behavior (e.g., checking

Facebook) before getting a reward (e.g., likes on their post) is not fixed. “Variable” means the size of the reward (e.g., the number of likes) is not fixed, either. Platforms use intermittent variable rewards both in how they prioritize content (“I might be rewarded with an interesting video if I keep scrolling.”) and in their social feedback systems (“I might be rewarded with a like or comment if I check Facebook.”). The use of intermittent variable rewards has caused some people to liken social media to slot machines. Id. at 326–27; Von Tristan Harris, The Slot Machine in Your Pocket, Spiegel Int’l (July 27, 2016) (Doc. 37-2).2 Platforms also use engagement-based algorithms that rely on the wealth of data platforms collect about and from their users: “by monitoring the amount of time particular

kinds of content keep the particular user engaged with the platform,” the platform can select and order content in a way that is likely to keep that user online for longer by, for example, waiting to show the most interesting content until right before it predicts the user will close the app. Bhargava & Velasquez, supra, at 334. This data also allows platforms to target advertisements based on user interest. More data about users and more computing power means platforms’ effectiveness in keeping users online is likely to continue improving. Social media is particularly fraught for young people, who have less developed impulse control systems and are especially sensitive to social feedback, peer pressure, and the negative cognitive effects of insufficient sleep. Potential Risks of Content,

Features, and Functions: The Science of How Social Media Affects Youth, Am. Psych. Ass’n (Apr. 2024) (Doc. 37-12).3 The American Psychological Association and the Surgeon General have both issued health advisories regarding minors’ social media use. Id.; U.S. Surgeon General, Social Media and Youth Mental Health (2023).4

2 https://www.spiegel.de/international/zeitgeist/smartphone-addiction-is-part-of-the- design-a-1104237.html [https://perma.cc/2MM8-NB4G].

3 https://www.apa.org/topics/social-media-internet/youth-social-media-2024 [https://perma.cc/A7GV-QRTL].

4 https://www.hhs.gov/sites/default/files/sg-youth-mental-health-social-media- advisory.pdf [https://perma.cc/3HU9-S6ZA]. Defendants say that Act 901 is necessary because social media platforms “hold a vast amount of power over Arkansans” but “have refused to exercise that power responsibly.” (Doc. 36, p. 4). Indeed, social media is used by billions of people worldwide, and millions in the United States. The average user spends hours a day on social media.

Vaishnavi Jahagirdar et al., Assessment of the Impact of Social Media Addiction on Psychosocial Behavior Like Depression, Stress, and Anxiety in Working Professionals, BMC Psych., 2024, at 2 (Doc. 37-4). And “a tiny proportion of Americans—between 5 and 10%—show addicted behaviors and use [social media] obsessively.” Id. In addition to social media addiction itself, Act 901 is also meant to address eating disorders, suicide, and drug use brought on by social media. Social media use has been linked to negative body image: Meta’s own research found that “[t]hirty-two percent of teen girls said that when they feel bad about their bodies, Instagram made them feel worse.” Georgia Wells, Jeff Horwitz & Deepa Seetharaman, Facebook Knows Instagram Is Toxic for Teen Girls, Company Documents Show, Wall St.

J. (Sept. 14, 2021) (Doc. 37-7);5 Zubair, supra, at 2 (“This appearance-related preoccupation inflicted by social media has been found to be directly proportional to its usage . . . .”). Research also indicates that “[e]xtended use of social networking sites . . . may be associated with negative symptoms of depression, anxiety, and stress.” Id.; Ujala Zubair et al., Link Between Excessive Social Media Use and Psychiatric Disorders, 85 Annals of Med. & Surgery 875, 875 (2023) (Doc. 37-19). And excessive internet use has been linked

5 https://www.wsj.com/tech/personal-tech/facebook-knows-instagram-is-toxic-for-teen- girls-company-documents-show-11631620739. to increased suicidal ideation: “Several studies have shown that those who manifest an internet addiction, including a social media addiction, are more likely than others to have suicidal ideation.” Bhargava & Velasquez, supra, at 330 (citations omitted); but see Zubair, supra, at 877 (“Social media usage can increase the probability of self-injury by

conveying different methods of self-mutilation, a sense of competition, and contagion, but these sites are also an important source of online emotional support.”). Finally, Defendants offer one report which found that TikTok served accounts registered to minors “videos about drug use, references to cocaine and meth addiction, and promotional videos for online sales of drug products and paraphernalia,” and another study which observed a link between alcohol overuse and “watching a higher number of alcohol-related posts.” Rob Barry et al., How TikTok Serves Up Sex and Drug Videos to Minors, Wall St. J. (Sept. 8, 2021) (Doc. 37-6);6 Zubair, supra, at 877. For its part, NetChoice does not deny that its members’ platforms are designed to maximize the time users spend on them. But it does assert that its “members expend

significant resources curating the content on their services to ensure that it is appropriate for adults and teens alike.” (Doc. 24, p. 6). NetChoice’s members have policies “that restrict content related to illicit drugs.” (Doc. 26, ¶ 9). They also regulate content that promotes suicide, self-harm, or eating disorders. Id. ¶ 10. And they purport to provide additional protections for minor users by “age gating” certain sensitive content that is not prohibited outright. See Doc. 24, p. 7. Enforcing these policies is expensive, and prohibited content has a way of getting through the cracks. Meta, for example, has spent over $20 billion on “safety and security

6 https://www.wsj.com/tech/tiktok-algorithm-sex-drugs-minors-11631052944. since 2016” and employs over 40,000 people in safety and security roles. (Doc. 28, ¶ 32). It uses “both automated and human review to detect violations” because, with billions of pieces of user-generated content coming in every day, “it is not possible to manually review every piece of user-generated content that potentially violates Meta’s policies.” Id.

¶ 44. In January through March of this year, YouTube removed nearly 3 million channels, over 8.5 million videos, and over 1.2 billion comments. (Doc. 25-15). The bulk of these removals are effectuated through YouTube’s automated system. Id. As this Court has previously found, there is no doubt that unfettered social media access can harm minor users. And there is evidence to support the conclusion reached by the Arkansas legislature that unfettered social media access can also harm adult users. But under existing First Amendment jurisprudence, the scope of that harm relative to the efforts social media platforms take to prevent it is largely beside the point. See United States v. Stevens, 559 U.S. 460, 470 (2010) (“The First Amendment's guarantee of free speech does not extend only to categories of speech that survive an ad hoc

balancing of relative social costs and benefits.”). II. LEGAL STANDARD In determining whether to grant a motion for preliminary injunction, the Court must weigh the following four considerations: (1) the threat of irreparable harm to the moving party; (2) the movant's likelihood of success on the merits; (3) the balance between the harm to the movant if the injunction is denied and the harm to other party if the injunction is granted; and (4) the public interest. Dataphase Sys., Inc. v. CL Sys., Inc., 640 F.2d 109, 114 (8th Cir. 1981). When a plaintiff seeks to “enjoin the implementation of a duly enacted state statute,” they must meet a more rigorous standard on the second prong by showing that they are “likely to prevail on the merits.” Planned Parenthood Minn., N.D., S.D. v. Rounds, 530 F.3d 724, 732–33 (8th Cir. 2008) (en banc). “While no single factor is determinative, the probability of success factor is the most significant.” Kodiak Oil & Gas (USA) Inc. v. Burr, 932 F.3d 1125, 1133 (8th Cir. 2019) (citation and quotation omitted). In particular, “[w]hen a Plaintiff has shown a likely violation of his or her First Amendment

rights, the other requirements for obtaining a preliminary injunction are generally deemed to have been satisfied.” Phelps-Roper v. Troutman, 662 F.3d 485, 488 (8th Cir. 2011) (per curiam), vacated on reh'g on other grounds, 705 F.3d 845 (8th Cir. 2012). III. DISCUSSION A. Severability Before reaching NetChoice’s merits arguments, the Court addresses the parties’ severability arguments. The Court generally agrees with NetChoice that if Defendants want the Court to sever a statute, they must raise severability. But in this preliminary injunction posture, the severability question is bound up with the scope of relief question that Defendants did raise. See Doc. 36, pp. 31–32. The Court is powerless to enjoin provisions of state law for which no Defendant in this case is imbued with enforcement

authority. See Whole Woman's Health v. Jackson, 595 U.S. 30, 43–44 (2021) (“[U]nder traditional equitable principles, no court may ‘lawfully enjoin the world at large,’ or purport to enjoin challenged ‘laws themselves.’” (citations omitted)). Counsel for Defendants represents that § 1503(b) is limited to a private right of action by a parent or guardian and that the Attorney General does not have the authority to enforce this provision. States generally enjoy sovereign immunity in federal courts from suits brought by citizens of other states and their own citizens. . . . Ex parte Young permits some actions for prospective injunctive relief against state officials acting in their official capacities, so long as the official has ‘some connection with the enforcement’ of the challenged law. Bio Gen LLC v. Sanders, 142 F.4th 591, 604 (8th Cir. 2025) (quoting Ex parte Young, 209 U.S. 123, 157 (1908). Whatever the propriety of § 1503(b), the Court cannot, consistent with the Eleventh Amendment, enjoin it here, where Defendants disclaim enforcement authority.

Therefore, in the discussion below, the Court considers only the remaining provisions of Act 901. The Court assumes for purposes of this Motion that the Attorney General has authority to enforce § 1503(a), which imposes “[a] civil penalty not to exceed ten thousand dollars ($10,000) per violation” on “[a] social media platform that knowingly and willfully violates this section.” Per Defendants’ concessions (and the plain text of the statute), the portions of Act 901 that refer to “this section” apply only to the section in which they are codified, e.g., § 1502(b)’s safe harbor provision is limited to § 1502, and § 1503(d)’s carve-outs are limited to § 1503. B. Likelihood of Success on the Merits 1. First Amendment The First Amendment protects Americans against government action “abridging

the freedom of speech.” Social media users engage in protected First Amendment activity on social media platforms. Packingham, 582 U.S. at 105. And the platforms themselves engage in protected First Amendment activity when they make editorial choices “about what third-party speech to display and how to display it.” Moody, 603 U.S. at 716. NetChoice brings both facial and as-applied challenges to Act 901. A law is facially invalid under the First Amendment “if the law's unconstitutional applications substantially outweigh its constitutional ones.” Moody, 603 U.S. at 724. Defendants assert that NetChoice cannot succeed on its facial First Amendment challenge because Act 901 has many constitutional applications, although they failed to identify any. In Moody, the Supreme Court looked at two facial challenges to state laws that “limit[ed] [social media] platforms’ capacity to engage in content moderation” based on the states’ concern that platforms were censoring certain speech and speakers. 603 U.S. at 717, 722. The Court concluded that both lower courts had improperly applied the facial

challenge standard because they failed to consider “the full range of activities the laws cover, and measure the constitutional against the unconstitutional applications,” and instead only looked at the constitutionality of certain “heartland applications,” namely “Facebook’s News Feed and its ilk.” Id. at 724. Per Moody, “[t]he first step in the proper facial analysis is to assess the state laws’ scope. What activities, by what actors, do the laws prohibit or otherwise regulate?” Id. On the “what actors” question, the laws at issue in Moody covered more, and more diverse, actors than does Act 901. See id. at 720–21. There, the laws defined “social media platform” expansively, potentially reaching “an email provider like Gmail,” “an online marketplace like Etsy,” and “a payment service like Venmo.” Id. at 725. Here, Act

901 is limited to website or applications that, among other things, “[e]mploy[ ] features that allow a user to connect, follow, or establish a relationship with other users,” and “[g]enerate[ ] revenue primarily through user engagement, including without limitation through advertising, user data monetization, or premium content.” It therefore does not regulate e-commerce platforms like Uber, Etsy, or Venmo (which do not generate revenue primarily through user engagement) or email providers like Gmail (which does not allow users to connect, follow, or establish relationships with other users). Instead, Act 901 applies to social media platforms as commonly understood, Facebook and Instagram, TikTok, Reddit, Snapchat, Pinterest, X, and whatever new platforms may hit the scene in the future. As to “activities,” § 1502 regulates any “design, algorithm, or feature” on any regulated platform to the extent that that design, algorithm, or feature causes a prohibited

result. This capacious language does, as Defendants argue, cover many applications, some that implicate either the platform’s or the user’s speech, and some that do not. As far as the Court can tell, this includes the prototypical recommendation algorithms used to order Facebook’s News Feed and YouTube’s homepage, as well as filters, social feedback i.e., “likes,” search functions, direct messaging, endless scroll, privacy settings (or the lack thereof), data-collection practices, and even where platforms put the “delete account” button. There are probably other “designs, algorithms, or features” that neither the parties nor the Court have identified. In sum, the Act regulates pretty much everything a social media platform does. Section 1503 regulates “host[ing], promot[ing], shar[ing], or otherwise facilitat[ing]

the immediate connection between” a minor user and “online content promoting, or otherwise advocating, self-harm or suicide,” if, “following exposure to” that content, the minor “commits suicide or attempts to commit suicide that results in significant bodily or cognitive harm.”7 Section 1503 has carve-outs for “[d]isplaying content that is created and hosted entirely by a third party, including without limitation an advertisement managed by a third party and shared on the social media platform” and for conduct protected by the First Amendment or the Arkansas Constitution.

7 Act 901 defines a minor as “an individual under sixteen (16) years of age.” § 1501(a)(2). “The next order of business is to decide which of the laws’ applications violate the First Amendment, and to measure them against the rest.” Moody, 603 U.S. at 725. One important difference between this case and Moody is that the challenged laws in Moody restricted the removal or suppression of user speech, while Act 901 does the opposite,

restricting the dissemination or promotion of user speech. This raises different First Amendment concerns because forcing a platform to carry user speech it would rather remove burdens the platform’s speech, but not the user’s, while forcing a platform to suppress user speech it would rather promote burdens both the platform’s and the user’s speech. For example, the Moody Court suggested the challenged laws might constitutionally be applied to direct messaging services on platforms to the extent that such services do not involve a platform’s editorial judgment. Moody, 603 U.S. at 725. But here, Act 901 would plainly be unconstitutional if it was applied to require platforms to censor users’ direct messages about, for example, marijuana use. Therefore, applications that do not involve the platform’s editorial discretion are still subject to First Amendment

scrutiny when they burden users’ speech.8 Starting with the likely constitutional applications, Defendants failed to identify any. They offer just two purportedly constitutional examples: “[A] particular notification method,” (Doc. 36, p. 13), which is likely protected speech. NetChoice v. Bonta, 761 F. Supp. 3d 1202, 1224–28 (N.D. Cal. 2024). And a Snapchat filter that encourages users to drive at dangerous speeds, (Doc. 36, p. 10), which is not covered by Act 901.

8 Under the facial challenge analysis, the Court must consider the constitutionality of all applications of the Act, not just those applications that implicate the particular plaintiff’s First Amendment rights. Outside Defendants’ examples, the Court can conceive of some applications of § 1502 that are likely constitutional because they involve platforms’ non-expressive conduct and impose minimal incidental burdens on speech. Such applications may include: appearance-altering filters; data collection practices; display features unrelated

to the ordering of content, e.g., infinite scroll, autoplay; and difficult to access parental controls or “delete account” buttons. See In re Soc. Media Adolescent Addiction/Pers. Inj. Prods. Liab. Litig., 702 F. Supp. 3d 809, 836 (N.D. Cal. 2023), motion to certify appeal denied, 2024 WL 1205486 (N.D. Cal. Feb. 2, 2024); Comput. & Commc'ns Indus. Ass'n v. Paxton, 747 F. Supp. 3d 1011, 1035–36 (W.D. Tex. 2024); see infra pp. 21–22. Section 1503, however, always premises liability on the content of the speech to which a user is exposed—“online content promoting, or otherwise advancing, self-harm or suicide”—and therefore has no application that is exempt from First Amendment scrutiny. In its more constitutionally suspect applications, § 1502 also implicates platform or

user speech by requiring platforms to change what speech they disseminate or how they prioritize that speech.9 These applications must therefore withstand heightened scrutiny under the First Amendment. Like § 1503, three of § 1502’s four prohibited results (drugs, eating disorders, and suicide) impose content-based restrictions on platforms’ editorial discretion and on users’ speech. And the fourth (social media addiction), even assuming it is content neutral, is not narrowly tailored to further the State’s interests.

9 The First Amendment protects a user’s right to disseminate their own speech and to receive the speech of others against government infringement. Sorrell v. IMS Health Inc., 564 U.S. 552, 568 (2011); Va. State Bd. of Pharmacy v. Va. Citizens Consumer Council, Inc., 425 U.S. 748, 756–57 & n.15 (1976). “‘Content-based laws—those that target speech based on its communicative content—are presumptively unconstitutional and may be justified only if’ they satisfy strict scrutiny. That standard requires that a law be ‘the least restrictive means of achieving a compelling state interest.’” Free Speech Coal., Inc. v. Paxton, 606 U.S. 461, 471 (2025)

(first quoting Reed v. Town of Gilbert, 576 U.S. 155, 163 (2015); and then quoting McCullen v. Coakley, 573 U.S. 464, 478 (2014)). Content-neutral laws are subject to “intermediate scrutiny.” Id. “Under that standard, a law will survive review ‘if it advances important governmental interests unrelated to the suppression of free speech and does not burden substantially more speech than necessary to further those interests.’” Id. (quoting Turner Broad. Sys., Inc. v. FCC, 520 U.S. 180, 189 (1997)). Section 1502’s first three prohibited results, and § 1503 as a whole, clearly impose content-based restrictions on speech. Section 1502 is susceptible to two possible interpretations, but either one would impose a content-based restriction. A “design, algorithm, or feature that . . . causes [a prohibited result]” could be read to include only

those designs, algorithms, or features that, by themselves and without user input, cause a prohibited result. For example, one could imagine that TikTok’s recommendation algorithm always prioritizes videos by BMI, with thinner users’ videos being ranked higher. This algorithm may, by itself, cause a user to develop an eating disorder. And § 1502 would forbid TikTok from making the editorial choice to prioritize thinner users—a content-, and perhaps even viewpoint-, based restriction on TikTok’s speech. But a “design, algorithm, or feature that . . . causes a [prohibited result]” might also include a design, algorithm, or feature that the platform knows may cause a prohibited result in response to certain user input.10 For example, YouTube, like many platforms, has a search feature. The search feature in and of itself is unlikely to cause a prohibited result. But sometimes it’s a bad idea to show a user the content they search for. For example, if a user searches “how to tie a noose,” YouTube knows that displaying videos responsive

to that search risks causing the user to commit or attempt to commit suicide, so YouTube, of its own volition, displays a “You’re not alone” message, links for a suicide crisis line, and a note that “the following results may be about suicide or self-harm” with an option to “show anyway.”11 Can YouTube be held liable under Act 901 for offering the “show anyway” option? Taking it a step farther, YouTube generally makes recommendations on its homepage based on a user’s search history.12 A user who searched for cat videos in the past will be recommended cat videos in the future. But YouTube should probably know that recommending videos about nooses to a user who searched “how to tie a noose” in the past could cause that user to attempt suicide. If YouTube treats “how to tie a noose”

like any other search for purposes of its recommendation algorithm, it may be subject to liability under Act 901. To avoid that liability, YouTube cannot maintain its preferred recommendation algorithm and must instead incorporate the State’s editorial judgment

10 This second reading is, in the Court’s view, the better reading because Defendants specifically point to engagement-based algorithms as an example of the harmful algorithms sought to be regulated by Act 901, see Doc. 36, p. 5, but engagement-based algorithms are, as the name suggests, necessarily responsive to user engagement.

11 https://www.youtube.com/results?search_query=how+to+tie+a+noose [https://perma.cc/W8KM-LWSF].

12 The editorial judgments influencing YouTube’s homepage were a prototypical example of protected platform activity in Moody. 603 U.S. at 744. by treating a “how to tie a noose” search differently than a cat search because of the search’s content. Defendants argue that § 1502 is not content based because it imposes liability on platforms based on the result caused by a design, algorithm, or feature without regard to

the particular content that design, algorithm, or feature causes to be displayed. But certain types of content, like the noose example, are known to be associated with certain results. See Bhargava & Velasquez, supra, at p. 337 (“A former drug addict, for example, may begin to experience [drug] cravings when seeing drug paraphernalia or watching a movie with scenes of people using drugs.”); Teen Mental Health Deep Dive, Wall St. J. (Sept. 29, 2021) (Doc. 37-17, p. 44) (content about a user wanting to hurt or kill themselves makes teens who view that content feel the worst). When platforms know that this type of content is posted on their platforms, § 1502 requires them to exercise reasonable care to ensure that their designs, algorithms, and features do not promote it, thereby removing platforms’ ability to, for example, use content-agnostic recommendation algorithms. And

Defendants can’t help but admit as much: If a social media platform knows based on its research and development that implementing a particular feature will cause users to develop an eating disorder or commit suicide, then it cannot implement that feature— regardless of whether the feature is simply a different type of ‘like’ button, an AI-algorithm pushing certain types of content, or sending notifications after a certain time of night. (Doc. 39, pp. 19–20). A law that prohibits platforms from pushing “certain types of content” but allows them to push other types of content is a content-based law. Because content-based laws are “presumptively unconstitutional,” Defendants must show that, with respect to § 1503 and the first three prohibited results of § 1502, Act 901 is narrowly tailored to serve a compelling state interest. Reed, 576 U.S. at 163. Assuming that Arkansas has a compelling interest “in protecting Arkansans from . . . self- harm behaviors, such as drug use, eating disorders, and suicide,” see Doc. 36, p. 21, it must pursue those interests through the “least restrictive available means,” which should be “neither seriously underinclusive nor seriously overinclusive,” United States v. Playboy

Ent. Grp., Inc., 529 U.S. 803, 823 (2000); Brown v. Ent. Merchants Ass'n, 564 U.S. 786, 805 (2011). Defendants have failed to prove that these provisions are narrowly tailored to serving the State’s asserted interests. Section 1502 imposes liability on platforms for disseminating content in a way that the platform “should have known” would cause any Arkansas user to purchase a controlled substance,13 develop an eating disorder, or attempt suicide—even if the vast majority of Arkansas users exposed to that content or dissemination method would not respond in those manners. By imposing liability on a platform any time the protected speech by or on that platform results in a specified “harm” to a single Arkansas viewer that the platform should have anticipated, § 1502 impermissibly limits the online posting

and promotion of protected speech that is not harmful to most viewers. The State cannot force platforms to censor potentially sensitive, but protected, speech as to all users for the benefit of some subset of particularly susceptible users. Instead, the burden of avoiding that speech should “normally fall[ ] upon the viewer to ‘avoid further bombardment of (his) sensibilities simply by averting (his) eyes.’” Erznoznik v. City of Jacksonville, 422 U.S. 205, 210–11 (1975) (quoting Cohen v. California, 403 U.S. 15, 21 (1971)). Section 1502 is substantially overinclusive.

13 Notably, “controlled substances” under Arkansas law are not limited to illicit drugs. See Ark. Code Ann. § 5-64-210 (tramadol); id. § 5-64-212 (ephedrine, pseudoephedrine). Section 1503—at least in its public enforcement provision—has less of an over- inclusivity problem than § 1502 because it requires that the platform “knowingly and willfully” facilitate a minor user’s exposure to online content promoting or advancing self- harm or suicide. Defendants assert that § 1503 is further narrowed by the carve-out from

platform liability for “[d]isplaying content that is created and hosted entirely by a third party, including without limitation an advertisement managed by a third party and shared on the social media platform,” which Defendants read as ensuring that platforms are never liable for mere failure to remove harmful content. § 1503(d)(1); Doc. 36, p. 17.14 Defendants have offered an interpretation of this carve-out that the Court finds less than persuasive but, as NetChoice argues, ultimately beside the point. Defendants assert that, whatever § 1503(d)(1) means, platforms are liable “when they push or promote certain content.” (Doc. 44, p. 5). But platforms’ decisions about what content to “prioritize” reflect protected editorial judgment, so even under Defendants’ interpretation, § 1503 burdens platforms’ speech based on its content.

Defendants have not met their burden to show that § 1503 is the least restrictive means of accomplishing the State’s interest in preventing suicide. “When plaintiffs challenge a content-based speech restriction, the burden is on the Government to prove that the proposed alternatives will not be as effective as the challenged statute.” Ashcroft v. Am. C.L. Union, 542 U.S. 656, 665 (2004). NetChoice proposes the less restrictive alternative of enforcing Arkansas’s existing criminal prohibition on “encouraging the

14 Defendants also argue that the carve-out for constitutionally protected conduct saves § 1503, but, considering that “content promoting, or otherwise advancing, self-harm or suicide” is protected, the Court finds that this clause is “absolutely inconsistent with the provisions of the act.” Atl. Richfield Co. v. Christian, 590 U.S. 1, 23 (2020) (citation omitted). suicide of another person.” Ark. Code § 5-10-107. Defendants fail to answer this proposal with their own argument or evidence for why enforcing this prohibition would not be as effective as § 1503 in preventing suicide. On the other side, both operative provisions of Act 901 are substantially

underinclusive because they do not restrict “identical content that is communicated through other media.” (Doc. 24, p. 19). Thus, 13 Reasons Why, a TV show which graphically depicts a suicide, can remain on Netflix, but YouTube could be liable for showing a user clips of the same. Similarly, Netflix can glorify thin bodies and cast exclusively thin actresses, but Instagram may be liable if it promotes those actresses’ posts thereby causing a user to develop an eating disorder. Defendants respond that these non-social media platforms do not have “the same type of emerging research and whistleblowers indicating that [they] are causing the same widespread harm that social media is causing.” (Doc. 39, pp. 22–23). But Defendants’ own evidence indicates that consumption of other types of digital media also risks the prohibited results. Bhargava &

Velasquez, supra, at 330 (“[P]ersons who have any kind of internet addiction not only think of suicide but also have significantly higher rates of planning and of actually attempting suicide.”); id. at 337 (“A former drug addict, for example, may begin to experience such cravings when seeing drug paraphernalia or watching a movie with scenes of people using drugs.”). “Underinclusivity creates a First Amendment concern when,” as in this case, “the State regulates one aspect of a problem while declining to regulate a different aspect of the problem that affects its stated interest in a comparable way.” Williams-Yulee v. Fla. Bar, 575 U.S. 433, 451 (2015). This underinclusivity is especially concerning because Act 901 does not generally bar minors’ access to speech promoting suicide or tending to cause a prohibited result; instead, it limits only their access to forums in which to discuss—rather than merely view—this speech. See Comput. & Commc'ns Indus. Ass'n, 747 F. Supp. 3d at 1037–38 (“A teenager can read Peter Singer advocate for physician-

assisted suicide in Practical Ethics on Google Books but cannot watch his lectures on YouTube or potentially even review the same book on Goodreads.”). These unregulated activities affect Arkansas’s stated interests in a comparable, if not identical way. Act 901 is therefore concerningly underinclusive. Defendants have failed to establish that §§ 1502(a)(1)–(3) and 1503 are narrowly tailored to achieving the State’s asserted interests in protecting Arkansans from drug use, eating disorders, and suicide. These provisions of the Act are likely unconstitutional. Finally, § 1502(a)(4) prohibits designs, algorithms, or features that cause a user to “[d]evelop or sustain an addiction to the social media platform.” This provision cannot survive intermediate scrutiny, so the Court need not determine whether it is content based

and assumes that it is content neutral. A content-neutral law burdening speech is consistent with the First Amendment “if it advances important governmental interests unrelated to the suppression of free speech and does not burden substantially more speech than necessary to further those interests.” Free Speech Coal., 606 U.S. at 471. Defendants assert that “Arkansas has a compelling interest in protecting Arkansans from social media addiction.” (Doc. 36, p. 21). Generally, state police powers include the authority to regulate in the interest of public health, including by restricting or prohibiting addictive substances or activities. See Minnesota ex rel. Whipple v. Martinson, 256 U.S. 41, 45 (1921) (“There can be no question of the authority of the state in the exercise of its police power to regulate the administration, sale, prescription and use of dangerous and habit-forming drugs . . . .”); Hawkeye Commodity Promotions, Inc. v. Vilsack, 486 F.3d 430, 439 (8th Cir. 2007) (“A state's police power encompasses controlling gambling, even to the point of abolishing a particular lottery game.”). But the

recent emergence of “addictive” technologies that disseminate speech puts states’ traditional police powers in safeguarding the public health in significant tension with the First Amendment’s guarantees. Buchanan v. Warley, 245 U.S. 60, 74 (1917) (“[I]t is equally well established that the police power, broad as it is, cannot justify the passage of a law or ordinance which runs counter to the limitations of the federal Constitution.”). The Supreme Court has recognized “that there is a compelling interest in protecting the . . . psychological well-being of minors.” Sable Commc'ns of California, Inc. v. F.C.C., 492 U.S. 115, 126 (1989). It has therefore “sustained legislation aimed at protecting the . . . emotional well-being of youth even when the laws have operated in the sensitive area of constitutionally protected rights.” New York v. Ferber, 458 U.S. 747, 757

(1982). It is significantly less clear that states are permitted to regulate in these sensitive areas to protect the psychological well-being of adults. Cf. Boos v. Barry, 485 U.S. 312, 321 (1988) (holding that “psychological damage” is not a “secondary effect” of speech that can render a facially content-based statute content neutral). As stated above, some designs or features that may make a platform “addictive” do not implicate platforms’ editorial discretion, and prohibiting or restricting these features would imposes little, if any, burden on users’ speech. For example, § 1502 might prohibit “infinite scroll.” The use of infinite scroll by platforms can hardly be considered expressive, and replacing infinite scroll with click to load more or pagination would still provide users easy access to the same amount of speech as infinite scroll. This change may help reduce compulsive use by providing stopping cues to users, furthering the State’s asserted interest without substantially burdening users’ speech. See Bhargava & Velasquez, supra, at 327. The same analysis applies to autoplay. Section 1502’s addiction

component is likely constitutional as applied to both these features. However, the more pressing addiction driver, the use of intermittent variable rewards, is achieved through designs, algorithms, and features that present more difficult questions. Social feedback features, e.g., likes, shares, or comments, by their very nature provide randomly timed rewards because they depend upon the independent action of another user, not the platform. Prohibiting this social feedback because it causes addiction in some users burdens the speech of the many other users for whom this feedback is not addictive. Platforms also use algorithms to order the content they display, and many of these algorithms incidentally or intentionally incorporate intermittent variable rewards.15 In

general, an algorithm is simply a “systematic procedure that produces—in a finite number of steps—the answer to a question or the solution of a problem.” Algorithm, Britannica.com.16 In the social media context, an algorithm is “a set of instructions or rules

15 For example, even the simple reverse chronological algorithm may create an intermittent variable reward schedule for users because, as a result of the independent actions of the accounts a user follows, the posts displayed will naturally vary in how interesting or “rewarding” they are.

16 https://www.britannica.com/science/algorithm [https://perma.cc/EWJ4-PHQ6]. that a computer application . . . uses to decide what content . . . should be shown to a particular user” and in what order. Algorithm, Dictionary.Cambridge.com.17 In a world where billions of pieces of content are posted on social media every day, social media would be functionally useless as a “vast democratic forum[ ]” if platforms

were not allowed to use any algorithm—any system—for selecting and ordering content to display to users. Reno v. Am. C.L. Union, 521 U.S. 844, 868 (1997). If a social media platform was a library, banning algorithms would be roughly equivalent to requiring books be placed on shelves at random. Such a prohibition would burden users’ (or library patrons’) First Amendment rights by making it significantly more difficult to access speech a user wishes to receive, so a state probably could not constitutionally ban algorithms for the organization of speech (on social media or elsewhere) altogether. But § 1502 does not prohibit all algorithms, only those that a “platform knows, or should have known through the exercise of reasonable care, cause[ ] a user to . . . develop or sustain an addiction to the social media platform.” NetChoice says § 1502 “forces

covered services to restrict as to all users—minor or adult—anything that could have a forbidden effect [here, addiction] on any users—again, minor or adult.” (Doc. 24, p. 18). And Defendants do not dispute this characterization. Accordingly, Defendants cannot claim that the Act is narrowly tailored. The tailoring requirement does not simply guard against an impermissible desire to censor. The government may attempt to suppress speech not only because it disagrees with the message being expressed, but also for mere convenience. Where certain speech is associated with particular problems, silencing the speech is sometimes the path of least resistance. But by demanding a close fit between ends and means, the tailoring requirement

17 https://dictionary.cambridge.org/dictionary/english/algorithm [https://perma.cc/4E7W- WQ8X]. prevents the government from too readily “sacrific[ing] speech for efficiency.” McCullen, 573 U.S. at 486 (alteration in original) (quoting Riley v. Nat’l Fed’n of Blind of N. C., Inc., 487 U.S. 781, 795 (1988)). Even without considering the First Amendment interests of the platforms, Act 901 impermissibly “regulate[s] expression in such a manner that a substantial portion of the burden on speech does not serve to advance its goals” by denying all users access to “a distinctive expressive product” they wish to receive because of the threat of addiction in any subset of users, no matter how small. Id.; Moody, 603 U.S. at 731. And the mens rea requirement does not help: a platform may know that an algorithm is addictive in some habitual users and still wish to give more temperate

users access to algorithmically curated content, but § 1502 directs that the platform “shall not use” the algorithm at all. Assuming the State has an important interest in preventing social media addiction in children, “that interest does not justify an unnecessarily broad suppression of speech addressed to adults. As [the Supreme Court] ha[s] explained, the Government may not ‘reduc[e] the adult population . . . to . . . only what is fit for children.’” Reno, 521 U.S. at 875 (ellipses and third alteration in original) (quoting Denver Area Educ. Telecomm. Consortium, Inc. v. FCC, 518 U.S. 727, 759 (1996)). Likewise, assuming the State also has an important interest in preventing social media addiction in adults, that interest does not justify reducing the more temperate user to only those algorithms that are suitable for

the habitual user.18 With respect to the addiction provision, § 1502 is not sufficiently tailored.

18 It is possible that some algorithms—especially purely or primarily time-tracking engagement-based algorithms like TikTok’s—could be prohibited with minimal burden on the more temperate user, who could instead access the same content of interest by Act 901 prohibits a substantial amount of protected speech relative to its plainly legitimate sweep. Because NetChoice is likely to succeed on its facial challenge, the Court does not consider Defendants’ arguments regarding NetChoice’s ability to bring as- applied challenges on behalf of its members.

2. Void for Vagueness NetChoice also challenges Act 901 on the ground that it is so vague it violates due process. It is a basic principle of due process that an enactment is void for vagueness if its prohibitions are not clearly defined. Vague laws offend several important values. First, because we assume that man is free to steer between lawful and unlawful conduct, we insist that laws give the person of ordinary intelligence a reasonable opportunity to know what is prohibited, so that he may act accordingly. Vague laws may trap the innocent by not providing fair warning. Second, if arbitrary and discriminatory enforcement is to be prevented, laws must provide explicit standards for those who apply them. Grayned v. City of Rockford, 408 U.S. 104, 108 (1972). “The degree of vagueness that the Constitution tolerates—as well as the relative importance of fair notice and fair enforcement—depends in part on the nature of the enactment.” Vill. of Hoffman Ests. v. Flipside, Hoffman Ests., Inc., 455 U.S. 489, 498 (1982). “[E]conomic regulation is subject to a less strict vagueness test” as are “enactments with civil rather than criminal penalties because the consequences of imprecision are qualitatively less severe.” Id. at 498–99. “Finally, perhaps the most

following specific users or topic. Some algorithmic features, like mixing in content the algorithm identifies as less interesting to maintain the intermittent variable reward schedule, could also be prohibited without burdening users’ access to speech, although such prohibitions may burden platforms’ speech to the extent the less interesting content reflects platforms’ editorial preferences. And requiring that users opt in to addictive features, rather than forbidding them altogether, would also be less burdensome on users’ speech. important factor affecting the clarity that the Constitution demands of a law is whether it threatens to inhibit the exercise of constitutionally protected rights. If, for example, the law interferes with the right of free speech or of association, a more stringent vagueness test should apply.” Id. at 499.

Act 901 must meet this more stringent vagueness test because it both permits criminal penalties, see Ark. Code Ann. § 4-88-103, and interferes with the right of free speech, see infra section III.B.1. “When speech is involved, rigorous adherence to those requirements is necessary to ensure that ambiguity does not chill protected speech.” FCC v. Fox Television Stations, Inc., 567 U.S. 239, 253–54 (2012). Indefinite statutes “abut[ting] upon sensitive areas of basic First Amendment freedoms” may force the entities regulated “to ‘steer far wider of the unlawful zone’ than if the boundaries of the forbidden areas were clearly marked.” Baggett v. Bullitt, 377 U.S. 360, 372 (1964) (quoting Speiser v. Randall, 357 U.S. 513, 526 (1958)). “Free speech may not be so inhibited.” Id. Section 1502 imposes liability when a platform uses a “design algorithm, or

feature” that it “knows, or should know through the exercise of reasonable care, causes a user to: (1) Purchase a controlled substance; (2) Develop an eating disorder; (3) Commit or attempt to commit suicide; or (4) Develop or sustain an addiction to the social media platform.” A law may not subject the exercise of a constitutional right (like speech) to an unascertainable standard of conduct. The Supreme Court has therefore consistently struck down laws that restricted otherwise protected speech based solely on its impact on the feelings and conduct of others. For example in Coates v. City of Cincinnati, 402 U.S. 611 (1971), the Supreme Court struck down an ordinance “mak[ing] it a criminal offense for ‘three or more persons to assemble on any of the sidewalks and there conduct themselves in a manner annoying to persons passing by.” Id. (cleaned up). The Court noted that “[t]he word ‘annoying’ is a widely used and well understood word; it is not necessary to guess its meaning.” Id. at 612. But the ordinance was nonetheless

unconstitutionally vague because it subjects the exercise of the right of assembly to an unascertainable standard . . . . Conduct that annoys some people does not annoy others. Thus, the ordinance is vague, not in the sense that it requires a person to conform his conduct to an imprecise but comprehensible normative standard, but rather in the sense that no standard of conduct is specified at all. Id. at 614. Similarly, the problem with § 1502 is not so much that its prohibited results are vague, but that it fails to specify a standard of conduct to which platforms can conform and its violation entirely depends upon the sensitivities of some unspecified user and a judge or jury’s determination about what the platform “should have known” about those sensitivities. Such a law is unconstitutionally vague. See also Winters v. New York, 333 U.S. 507, 519 (1948) (holding invalid law which “propose[d] to punish the printing and circulation of publications that courts or juries may think influence generally persons to commit crime of violence against the person”). Defendants assert that any uncertainty in the Act’s proscription is ameliorated by the Act’s knowledge requirements. (Doc. 36, p. 24). But § 1502 does not have a knowledge requirement; it has a “knows, or should have known through the exercise of reasonable care” requirement.19 “[S]hould have known through the exercise of

19 It is true that criminal penalties are limited to knowing and willful violations, Ark. Code Ann. § 4-88-103, but civil enforcement actions are not, id. § 104. reasonable care” is an objective negligence standard. Counterman v. Colorado, 600 U.S. 66, 79 n.5 (2023). In the sensitive First Amendment context, objective standards like this one do not tend to decrease the threat posed by vague laws. Cf. id. at 78 (“An objective standard, turning only on how reasonable observers would construe a statement in

context, would make people give threats ‘a wide berth.’” (citation omitted)); id. at 88–89 (Sotomayor, J., concurring in part and concurring in the judgment) (noting potential for bias in a jury’s determination of what constitutes a threat). There is no narrowing construction the Court can give to § 1502 that would be consistent with its text and would render it constitutional—Defendants don’t even bother to suggest one. (Doc. 36, pp. 24– 25. Section 1502 is likely void for vagueness. Section 1503, on the other hand, does have a subjective mens rea requirement, knowing and willful. Under Arkansas law, willfulness generally requires something more than mere knowledge of the act itself. Instead, a person willfully violates a law when they knowingly act or fail to act with knowledge or in conscious disregard of the risks or

wrongfulness of that act. Sw. Bell Tel. Co. v. Ark. Pub. Serv. Comm'n, 68 Ark. App. 148, 157–58 (1999). First Amendment issues aside, § 1503(a) does not require platforms to engage in the kind of speculative predictive judgments about what content is likely to cause a prohibited result in a user; it applies to specific conduct and content (hosting, promoting, sharing, or otherwise facilitating the immediate connection between a user and online content promoting, or otherwise advancing, self-harm or suicide); and its mens rea provision does, as Defendants argue, reduce the threat of vagueness. Section 1503(a) is likely not void for vagueness. 3. Section 230 Preemption Finally, NetChoice argues that Act 901 is preempted by § 230 of the Communications Decency Act, 47 U.S.C. § 230. Section 230 provides 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.” Id. § 230(c)(1). An

“interactive computer service” is “any information service, system, or access software provider that provides or enables computer access by multiple users to a computer server, including specifically a service or system that provides access to the Internet.” Id. § 230(f)(2). NetChoice’s members are providers of interactive computer services. An “information content provider” is “any person or entity that is responsible, in whole or in part, for the creation or development of information provided through the Internet or any other interactive computer service.” Id. § 230(f)(3). Section 230 expressly preempts contrary state laws: “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.” Id. § 230(e)(3).

Section 230 “immunizes providers of interactive computer services against liability arising from content created by third parties.” E. Coast Test Prep LLC v. Allnurses.com, Inc., 971 F.3d 747, 752 (8th Cir. 2020) (quoting Fair Hous. Council v. Roommates.com, LLC, 521 F.3d 1157, 1162 (9th Cir. 2008) (en banc) (footnote omitted)). “Congress thus established a general rule that providers of interactive computer services are liable only for speech that is properly attributable to them.” Johnson v. Arden, 614 F.3d 785, 791 (8th Cir. 2010) (quoting Nemet Chevrolet, Ltd. v. Consumeraffairs.com, Inc., 591 F.3d 250, 254 (4th Cir. 2009)). When is a third party’s speech properly attributable to the interactive computer service on which it is posted? Well, it’s hard to say. See Gonzalez v. Google LLC, 598 U.S. 617 (2023). Defendants assert that platforms can be held liable whenever they engage in their own expressive activity, see Doc. 36, p. 9 (citing Anderson v. TikTok, Inc., 116 F.4th 180, 183 (3d Cir. 2024)), which, under the Supreme Court’s decision in Moody, would permit liability for platforms’ editorial decisions, see 603 U.S. at 717.

This Court respectfully disagrees with this argument and the Third Circuit decision on which it relies. This argument “presuppose[s] that editorial decisions cannot be both an expression of a publisher's point of view (protected under the First Amendment) and a publication of a third-party's content (protected under Section 230).” Doe (K.B.) v. Backpage.com, LLC, 768 F. Supp. 3d 1057, 1063 (N.D. Cal. 2025). But “the undisputed core of Section 230 immunity protects a website's moderation decisions, in its role as a ‘publisher,’ about which third-party content to remove and which to permit”—decisions which also receive First Amendment protection. Id.; § 230(b)(4)(“It is the policy of the United States . . . to remove disincentives for the development and utilization of blocking and filtering technologies . . . .”); Moody, 603 U.S. at 718. Immunizing platforms from

being treated as publishers only when they are not acting like publishers would nullify Section 230 and contradict its plain text. NetChoice, on the other hand, asserts that “[a]ll of Act 901’s liability-creating provisions . . . impermissibly make[ ] online services ‘liable for information originating with third-party users of the service.’” (Doc. 24, pp. 24–25 (quoting Arden, 614 F.3d at 792). But this argument also goes too far. Outside the First Amendment context, a facial challenge can succeed only if the challenger “establish[es] that no set of circumstances exists under which the Act would be valid.” United States v. Salerno, 481 U.S. 739, 745 (1987); see Anderson v. Edwards, 514 U.S. 143, 155 n.6 (1995) (applying Salerno to facial preemption challenge). NetChoice is not likely to succeed on its facial preemption challenge because some applications of Act 901 are consistent with Section 230. In some potential

applications, e.g., designs that make it difficult to delete one’s account, liability is not premised on a platform’s editorial decisions about third-party content at all. In others, e.g., appearance-altering filters, Snapchat streaks, the platform’s own posts, the platform itself is acting as an information content provider, and Section 230 does not immunize information content providers from liability for their own content. In re Soc. Media Adolescent Addiction, 702 F. Supp. 3d at 830; see Lemmon v. Snap, Inc., 995 F.3d 1085 (9th Cir. 2021). As such, the Court cannot conclude that Act 901 is preempted by Section 230 in all its applications. C. Irreparable Harm “When a plaintiff has shown a likely violation of his or her First Amendment rights,

the other requirements for obtaining a preliminary injunction are generally deemed to have been satisfied.” Minn. Citizens Concerned for Life, Inc. v. Swanson, 692 F.3d 864, 870 (8th Cir. 2012) (en banc) (citation omitted). “It is well-established that ‘[t]he loss of First Amendment freedoms, for even minimal periods of time, unquestionably constitutes irreparable injury.’” Powell v. Noble, 798 F.3d 690, 702 (8th Cir. 2015) (alteration in original) (quoting Elrod v. Burns, 427 U.S. 347, 373 (1976)). Because NetChoice has shown that it is likely to prevail on its First Amendment claim, its members are likely to suffer irreparable harm if Act 901 goes into effect. D. Balance of the Equities and the Public Interest Enjoining enforcement of a duly enacted statute does harm the government— “[uJnless that statute is unconstitutional.” Abbott v. Perez, 585 U.S. 579, 602-03 & n.17 (2018). “[I]t is always in the public interest to protect constitutional rights” and “[t]he balance of equities, too, generally favors the constitutionally-protected freedom of expression.” Phelps-Roper v. Nixon, 545 F.3d 685, 690 (8th Cir. 2008), overruled on other grounds by Phelps-Roper v. City of Manchester, 697 F.3d 678 (8th Cir. 2012) (en banc). Where, as here, the plaintiff has shown a likely First Amendment violation, any harm to the government posed by a preliminary injunction “cannot be held sufficient to overcome the public's interest in protecting freedom of expression under the First Amendment.” Rodgers v. Bryant, 942 F.3d 451, 458 (8th Cir. 2019). IV. CONCLUSION For these reasons, NetChoice’s Motion for Preliminary Injunction (Doc. 23) is GRANTED. Defendants and their officers, agents, affiliates, subsidiaries, servants, employees, successors, and all other persons or entities in active concert or privity or participation with them, are hereby PRELIMINARILY ENJOINED under Federal Rule of Civil Procedure 65(a) from enforcing the provisions of Act 901 for which they have enforcement authority against NetChoice’s members, pending final disposition of the issues on the merits. IT IS SO ORDERED on this 15" day of December, 2025.

CHIEF UNITED STATES DISTRICT JUDGE