“Segregate-and-Suppress:” A Solution in Search of a Solution
Viv Daniel
I. Introduction
The following paper is an analysis of Eric Goldman’s 2025 article published in the Stanford Technology Law Review, The “Segregate-and-Suppress” Approach to Regulating Child Safety Online.[1] Goldman’s article identifies an emerging legislative trend meant to protect children online, which he terms “segregate-and-suppress,” and argues that this legislative strategy is misguided because it damages privacy online, it is detrimental to the online information ecosystem, and it hurts many of the very children it was designed to protect. A segregate-and-suppress law is a law targeting publishers of content and information via websites and/or apps, which requires a publisher to distinguish between users on the basis of age, and to limit access to content for users deemed to be minors.[2]
This paper will begin by describing the problem that segregate-and-suppress was created to solve, to give context to the creation and implementation of these laws. Next, it will provide examples of the kinds of laws which fall under Goldman’s scrutiny and describe Goldman’s critiques of segregate-and-suppress and his alternative suggestions to it. Finally, this paper will evaluate the strength of Goldman’s arguments and proposed alternate solutions. This paper posits that, while Goldman’s argument is valuable to an honest debate of the topic, it would be strengthened by acknowledging the extent of the problem segregate-and-suppress is meant to solve, and by giving more consideration to the breadth of compromise-driven solutions available to alleviate threats to children’s safety online.
II. The Problem
An article identifying the segregate-and-suppress instinct in global legislation was bound to arrive, as addressing online threats to children has become one of the fastest-moving and most bipartisan priorities of lawmakers in recent years.[3] The desire to protect ourselves – and our children in particular – from the harms of online life has been a pervasive theme of the broader culture for years, but especially since the pandemic when society moved even more irrevocably online.[4] Even members of our most plugged-in generations feel the need to limit screentime. The sentiment from many young adults that they feel like they narrowly escaped the annihilation of their attention spans by being born pre-social media is common to find, ironically, on the internet. News pieces on the novelty of teens and twenty-somethings discovering serenity and community by giving up their smartphones break into the discourse, it seems, once a year.[5] On top of this, the 2025-2026 school year has seen a massive increase in the number of states and school districts banning non-emergency cellphone use during school hours.[6]
Of course, the understanding that the internet and smartphones can be dangerous to all demographics and can facilitate minors’ access to inappropriate content is not new. The dawn of the internet, and the advent of mass social media in particular, brought with it the novel ability of anyone with an internet connection to prey upon other users and to upload the most heinous acts imaginable – enough to disturb adult or child.[7] This problem necessitated content-moderation policies to guard against such behavior.[8] The breaking of the Cambridge Analytica scandal in 2016 was a watershed moment which helped lead to society’s more critical current view of social media, in which the unauthorized collection of user data and the power of such activities to influence international politics was put to light.[9] Increasingly, internet users are put-off by the creepiness of what scholar Shoshanna Zuboff calls, “surveillance capitalism,” and look to their lawmakers to do something about it.[10] The recent ascendance of AI and associated concerns about deepfakes, copyright infringement, worker replacement, environmental degradation, and AI chatbots’ tendency to exacerbate dangerous mental health crises, only serve to underscore the general sense that our devices harbor something sinister. However, for reasons which would require another dedicated paper to unpack, legislation to address these concerns can be slow-moving and patchwork in nature in the United States, although it does not fully lack momentum.
As Goldman points out in his paper, there is one area of tech-targeted legislation which does feel like it enjoys widespread and relatively uncomplicated support: Legislation aiming to protect children online. Whether doing so to protect or to control children, addressing perceived harms to children is something most legislators can get behind. This project is made easier by the fact that minors cannot vote out politicians who take away their access to Instagram. However, children’s easy access to content like online pornography and the exploitation of children in chatrooms have been open secrets since internet access went mainstream. Additionally, laws aiming to protect kids and their data online existed before the recent trend of what Goldman calls “segregate-and-suppress.” [11] Consequently, it is worth looking at which issues have tipped the scale lately and ushered in a new, more aggressive regime of regulation.
In the last decade or so, internet, social media, and device use have increasingly been linked to negative mental health outcomes, decreased performance in school, and radicalization among young people.
Perhaps most concerning of these trends, the potentially negative impact on mental health that device and social media use may have on youth, has been recognized for years. Studies have seemed to confirm what many people felt intuitively: That social media encourages us to compare our realities to the idealized lives of others – a line of thinking which children and teenagers are particularly hard-pressed to approach with a healthy perspective.[12] High profile cases of self-harm and suicide, such as the case of Molly Russell in 2017, have made parents painfully aware of the potential addictive capacity of social media, and the manipulative tactics being deployed against their children through it.[13] More recently, books like Jonathan Haidt’s The Anxious Generation have sparked deeper discussions about the potential dangers of giving young people unfettered access to smartphones and social media, even if there are strong and valid critiques about the weight Haidt gives other factors potentially contributing to youth anxiety.[14] As with all things tech, the rapid ascension of AI ubiquity has only intensified the perception that internet use is dangerous for children. AI chatbots’ concerning propensity for being agreeable and, at times, conspiratorial, has been tied to the suicides of multiple teenagers in the last few years, leading to calls for greater regulation of AI companies, and even some proposed legislation to ban minors from using AI chatbots.[15]
Concern around the impact of technology on academic performance has followed a similar path. Throughout the 2010s, as smartphone use increased rapidly among teenagers, school norms around phone use in class were not robust enough to handle the capacity for stimulation and distraction of modern devices – not to mention the temptation for students to spend their homework hours scrolling social media and short-circuiting their attention spans. Since the launch of ChatGPT in 2022, teachers have been inundated with AI-generated essays, forcing them to second-guess the authenticity of submitted assignments and to worry about whether students are developing the critical thinking skills they will need as adults.[16]
Finally comes the issue of radicalization, which is entwined at its base with concerns about mental wellness. Although algorithm-driven polarization and extremism has affected many different populations, it has especially become a concern for lonely and isolated young men. Some of the most destructive forces in modern society have been associated with communities of disaffected teenage boys and young men who have radicalized themselves through online forums.[17] The popularity of groups such as the incel movement has spawned violent attacks targeting women and the proliferation of regressive gender politics among young men.[18]
Amid all of this, there are those who have pushed for limiting minors’ access to the internet to further isolate marginalized youth, such as gay or transgender individuals, or to stop minors from having access to legitimate, age-appropriate content which would allow them to be informed about the world.[19] Still, the concerns listed above are only a brief sample of legitimate reasons to worry about the harms which children face in the online world. What to do about it, if anything, is subject to much debate. In his paper, Eric Goldman surveys the predominant legislative strategy which has emerged to remedy this set of crises – the “segregate-and-suppress” strategy – and finds that we have not yet landed on an appropriate solution.
III. What is “Segregate-and-Suppress”?
While the term “segregate-and-suppress” may be unfamiliar to Goldman’s readers, the type of law that the term describes is likely more recognizable. To fully understand what kinds of laws fall into the category of segregate-and-suppress laws, it is helpful to review some of Goldman’s illustrative examples. As stated above, a segregate-and-suppress law is a law targeting publishers of content and information via websites and/or apps, which requires a publisher to distinguish between users on the basis of age (this process is the “segregation” aspect), and to limit the access to content of users deemed to be minors (this is the “suppression” aspect).
Goldman references a number of segregate-and-suppress laws in his paper, but in introducing the concept, he specifically highlights three of them: H.B. 1181 from Texas, which requires publishers to enact segregate-and-suppress if at least one third of their content is deemed “harmful to minors;” California’s Protecting Our Kids from Social Media Addiction Act of 2024, which outlaws addictive feed mechanics for minors; and – most extreme of all – Australia’s Online Safety Amendment (Social Media Minimum Age) Act of 2024, which bans minors from social media entirely.[20] These examples effectively illustrate the heterogeneous nature of the laws captured by Goldman’s definitional umbrella. There is variety amongst these laws when it comes to which harms they are designed to combat, and to what extent/how publishers must limit minors’ access to online content under them. What these laws do all have in common, though, is the segregate-and-suppress scheme as articulated above.
IV. The Argument Against Segregate-and-Suppress
Goldman’s argument against segregate-and-suppress laws can be boiled down to three main objections: First, the age authentication processes needed to effectively segregate adults and minors online expose all users to heightened privacy and security risks. Second, he argues that a segregate-and-suppress model will detrimentally reshape the online information ecosystem. Finally, and somewhat counterintuitively, he says that these laws harm many of the minors they are meant to protect. Let us examine each argument in more detail.
1. The Privacy Dangers of Age Segregation Technology
The age segregation, or age verification technology that Goldman focuses on comes in three categories: Document review, visual inspection, and a third category of other strategies containing essentially everything else.[21]
a. Document Review
The document review solution for verifying users’ age involves internet users uploading a copy of their government-issued ID to a publisher so that the publisher can determine if the user is old enough to access restricted content when viewing their site. Document review seems logical, because it is analogous to the relatively innocuous age-verification we do in real life: i.e. requiring people to show official ID before purchasing alcohol, among other goods, or before entering adult-oriented spaces.
However, Goldman argues that what feels obvious and unobtrusive in the real world may nevertheless be infeasible and unsafe online. When one shows one’s ID in real life, the store or restaurant offering restricted products or services doesn’t usually make and store a copy of the ID.[22] Goldman notes that unlike in real life, IDs uploaded online will necessarily be stored by the publisher for a period of time, opening internet users up to harm if the publisher were to experience a security breach, and the risk that the government or other entities could track them more easily via the de-anonymizing of the internet.[23]
b. Visual Inspection
Visual inspection, an age-verification strategy involving uploading one’s photo for age review by a publisher or allowing a publisher to inspect one’s face via a device camera, similarly compromises internet users’ privacy in Goldman’s view. It is technically possible for facial inspection technology to merely view a user’s face without saving any data or associating a user’s appearance with other information about them.[24] However, Goldman points out that this clean, data-minimizing mode of inspection is almost never the case in reality.[25] Visual inspection technology, rather, is more often employed both to determine age and to contribute biometric data to users’ “persistent online identities,” further de-anonymizing the internet.[26] On top of this, Goldman notes that the benefits of visual inspection are not worth the risks because the technique is not very accurate.[27] This is especially true for people close to the age of majority, and thus relying on visual inspection to segregate users by age would result in many minors being allowed to view potentially harmful content, and many young adults being blocked from viewing content they are legally within their rights to seek out.
c. Other Strategies
Other age-verification strategies mentioned by Goldman he largely criticizes by arguing either that they are too ineffective or that they are just as invasive to privacy as the first two categories – if not more so. This third category includes methods such as capacity testing, device verification, data mining, social vouching, and the use of credit cards.[28] Goldman argues that strategies such as capacity testing (making users perform tasks associated with certain cognitive ages), device verification (associating the age of a device’s owner with the device in perpetuity), social vouching (allowing other users to confirm someone else’s age), and credit card use (assuming that anyone with a credit card must be an adult), are too easily gamed and are thus unreliable. The data mining option is particularly unacceptable because mining enough data on a user’s online activity to determine if they act like a child or an adult is clearly a massive invasion of privacy and goes against basic information privacy principles.
Goldman also points out that many age verification strategies are increasingly being outsourced to third parties who have an incentive to create and sell consumer profiles with whatever data they receive.[29] The final conclusion Goldman draws in his exploration of age verification seems clear: Effective strategies are too much of a privacy risk to be considered, and more privacy-friendly strategies are too ineffective to be considered, so we should probably avoid age segregation altogether.
2. The Degradation of the Online Information Ecosystem
Goldman’s next argument against segregate-and-suppress laws is that they will harm the quality and availability of information online by favoring large publishers and encouraging self-censorship. Large publishers will be favored because of the costs associated with complying with and thriving under segregate-and-suppress. Many segregate-and-suppress laws put the onus on publishers to make the call on what content is “harmful to children,” for instance, or what methods of age verification are adequate.[30] Without clear guidance, large, well-established publishers will be best equipped to afford the litigation costs associated with hashing out these issues. In this paradigm, many smaller publishers may choose to self-sensor rather than to take on liability under segregate-and-suppress laws.[31]
On top of this, segregate and suppress laws create technical authentication burdens which could have disastrous financial consequences for publishers, not only due to the costs associated with their upkeep, but also due to lost traffic to their sites.[32] Going through age-verification screening where an internet user might be asked to look into a camera or upload a file requires users to navigate through more interstitial pages before they can access the content they are seeking. Larger publishers are not only better able to maintain fast-loading websites with efficient age-verification tech, but they also benefit from having brand-recognition.[33] Lower wait times to access content, coupled with the confidence on users’ part that the content on the other side is worth a short wait, helps large publishers avoid the high “bounce rate”[34] and associated loss in revenue which will be suffered by less established competitors.[35]
3. Harm to Children
Goldman argues that segregate-and-suppress laws harm children because minors are not a monolith, so laws do not adequately account for competing interests among them, and they harm minors’ access to information.
Not all minors are susceptible to harm from the same content. Segregate-and-suppress laws do not generally distinguish between the access to information afforded to very young children versus that which is afforded to teenagers. Goldman does not necessarily think that “idiosyncratic suppression” (tailoring of suppression laws to different minor populations’ needs) is the right way to go either, because it would be more data-maximalist in practice and difficult to scale.[36] Assuming idiosyncratic suppression is impossible, Goldman points out that under “one-size-fits-all suppression,” many older minors could be disadvantaged by a lack of access to content, such as sex ed or mental health information, which would be beneficial and age-appropriate for them.[37] Additionally, Goldman notes that this dynamic can exist within children of the same age range and demographic. For instance, Goldman discusses the infamous 2019 survey which found that 21% of young girls who responded said that Instagram made them feel worse about themselves. Goldman points out that there is a flipside to this statistic which is cited much less often: 37% of young girls in the same study said that Instagram actually made them feel better about themselves. To craft laws which protect the 21% of girls harmed by Instagram’s content could mean sacrificing all of the benefits reaped by the 37%.[38]
The same way that limiting teenagers to content appropriate for five-year-olds circumscribes their intellectual exposure, the prevalence of culture war battles could cause a similar process to play out for much more nefarious reasons.[39] With the recent trend of book banning legislation in mind, it is not difficult to imagine that, where terms like “harmful to minors” and even “pornography” are ill-defined, they could be interpreted to include information about hot-button topics like gender identity.[40] Even where parental-consent carve-outs are made to allow minors access to potentially harmful content, this does not protect kids whose parents are unwilling to give the necessary consent.[41]
V. Policy Alternatives
After enumerating his objections to segregate-and-suppress as a solution to the dangers minors face online, Goldman lays out the policy solutions which he thinks should be pursued in its place. The article does not go into extensive detail about any of his policy proposals, but he touches on the following: Enacting digital literacy and citizenship education for children, training parents better to protect their kids online, funding more research on the online harms faced by different subsets of minors, implementing greater privacy protections for all internet users, enforcing existing laws which could help protect children online, and doing all of this while championing the First Amendment rights of minors.[42] Finally, Goldman notes that, if we do feel compelled to keep pursuing age segregation in online spaces, legislators should at least be expected to enumerate guidance to publishers about what methods to use.[43]
VI. Evaluation of the Article
The “Segregate-and-Suppress” Approach to Regulating Child Safety Online gives the privacy community language to describe a new legislative trend and appropriately urges that we pump the brakes where there are serious downsides to this trend. Goldman rightfully points out the drawbacks to the segregate-and-suppress approach which should give privacy advocates pause, especially where data maximization and censorship are at issue. It is also admirable that Goldman stands up for the speech rights of children upon whom we have imposed these imperfect legal solutions without consultation.
At the same time, Goldman’s overall point is weakened by some aspects of its presentation. To many who are concerned about minors’ safety online, Goldman’s policy solutions may read as underbaked. Most of them do not receive more than a few short paragraphs of explanation. Some are explained in only a couple of sentences. In a world containing an epidemic of radicalized teenage boys, AI chatbots which convince children to commit suicide, and algorithm-aggravated ADHD,[44] solutions which amount largely to ‘educate kids and families and fund more research’ do not quite match the urgency which sparked the enactment of segregate-and-suppress laws in the first place.
Additionally, one of Goldman’s most feasible and common-sense suggestions – enforcing laws which already exist – does not come with an elaboration on which laws he means. His citation in this section leads to a discussion of kids safety laws on the horizon at the time of publishing, such as KOSA and updates to COPPA.[45] However, considering that Goldman criticizes COPPA earlier in his article for incentivizing publishers to be willfully blind to whether children are using their sites,[46] it remains unclear which existing laws Goldman would like to see better enforced. Another of Goldman’s best and most pragmatic suggestions, asking legislatures to give guidance about segregation methods, may leave us vulnerable to overly specific statutes which fail to cover future verification methods, and legislators making technical recommendations completely outside their areas of expertise.
Goldman’s thesis similarly has the effect of understating potential harms to minors online. This is especially apparent in his discussion of the Instagram study. While it is true that the 37% of young girls who felt better about themselves after using the app may face restrictions on their access to it under some versions of segregate-and-suppress, it is not evident that this lost benefit is so equivalent to the mitigated harm to the 21%, as to cancel each other out. When a product like Instagram has the potential to exacerbate or induce mental health crises and eating disorders in some young users, one faces a heavy burden in arguing that the product also carries equivalent or greater benefits. Goldman does not attempt to meet this burden, other than by emphasizing that 37% is more than 21%.
One of the article’s strengths – the identification of “segregate-and-suppress” as a category – also causes some loss of clarity. Coining and applying this term allows Goldman to explain to the reader how much of the world has agreed upon a certain paradigm for legislating children’s safety online, no matter how different various jurisdictions’ politics and stated reasons for using the strategy. This framing is effective at demonstrating the risks – privacy and otherwise – shared by many different laws. However, Goldman’s use of “segregate-and-suppress” also paints with a broad brush. It seems clear that California restricting the use of addictive feed algorithms for kids is a wholly different issue (and perhaps one which weighs more favorably against any drawbacks) than Australia banning children from social media altogether. Although it is true that even moderated and well-meaning segregate-and-suppress laws can have negative consequences, and can even be enforced nefariously, Goldman’s near all-or-nothing approach to condemnation is one of the article’s weaknesses.
Goldman himself recognizes that he is capturing a wide variety of legislation within his definition of segregate-and-suppress. One aspect of the legislative strategy which he finds troubling is the fact that, “[t]here is no consistency in how the laws define minors, how the entities are supposed to determine who is a minor, what entities the laws regulate, and how the laws require those entities to restrict minors.”[47] However, the same diversity which makes these laws difficult to navigate for publishers also lends itself to more granular consideration of the merits of individual solutions. For instance, on the issue of censorship: Goldman is correct to be concerned about some segregate-and-suppress laws – especially those which explicitly regulate content – creating a more limited online ecosystem and damaging information access to both adult and minor members of marginalized communities. Still, as Goldman himself concedes, not all laws he defines as segregate-and-suppress operate this way and pose the same risk. Rather than taking aim at offensive content, laws can ban certain negatively impactful content presentation methods like infinite scroll, non-chronological timelines, or personalized algorithms.[48] Laws can also impose time-limits for minors on certain kinds of websites without banning access entirely.[49]
As Goldman writes in his overview of different kinds of segregate-and-suppress laws, such algorithm-targeting laws already exist, including in California.[50] While critiques of such laws on the grounds of technical difficulty and revenue loss may still be pertinent, Goldman’s aversion to all forms of segregate-and-suppress would feel more justified if he had more meaningfully engaged with the differences in these laws and the fact that not all of them are so quick to resort to censorship. In particular, Goldman’s argument about advantaging already powerful publishers falls flat where algorithm restrictions are concerned. As social media companies’ forceful defense against a recently successful landmark California lawsuit claiming algorithm addictiveness goes to show, challenging the mechanisms which keep people hooked on social media sites for hours at a time hits the largest tech giants where it hurts most.[51]
Goldman’s desire to fit many, quite different segregate-and-suppress laws into the same framework also causes inconsistencies in his argument due to the complexity of the issue. As noted above, Goldman dismisses as impossible any kind of scaled idiosyncratic suppression. When he later discusses potential solutions, he mentions the importance of further research and suggests that the government should collect data on and further investigate the impact of the internet on “niche subpopulations.”[52] Surely any policy suggestion to come out of such research would run into the same challenges which led Goldman to conclude that anything besides one-size-fits all problem solving is not workable. Similarly, Goldman advocates training parents to better teach their children about online safety.[53] This is a good suggestion, but it is undermined by Goldman’s own suggestion that many parents, no matter what you do, do not have their children’s best interests at heart.[54] Not to mention the fact that many low-income parents may not have the time to receive and impart such safety training. Why not educate parents where possible while also creating laws and regulations to fill the gaps? Finally, Goldman advocates for creating more effective privacy and online protection legislation which applies to everyone – not just to children.[55] This is also a good suggestion. Nevertheless, while the wheels of legislation are much more lubricated where children’s online harms are concerned, it may be prudent to support at least some of these efforts so that their success or failure might serve as a stepping stone to broader action.
Finally, Goldman’s argument against segregate-and-suppress laws turns the perfect into an enemy of the good. Although it takes some teasing out, Goldman structures his argument about privacy concerns around the idea that the most effective methods of age verification are too risky to accept and that the less risky methods are too ineffective to accept. This formulation, though, stacks the decks in Goldman’s favor to avoid considering that we could stick with segregate-and-suppress for now, but avoid using privacy intensive verification methods. Of course, age-gating and even device-verification are highly imperfect and not completely devoid of privacy concerns. Some data-lean forms of age-gating, such as allowing a user to click a button that says “yes, I am 18+” with no other verification, seem so obviously ineffectual that they are hardly addressed in Goldman’s article. Even when Goldman takes issue with both a method’s efficacy and its respect for privacy, such as his dismissal of idiosyncratic suppression, Goldman sometimes glosses over the nuances to frame his objections as obvious. In this particular example, Goldman creates a dichotomy between suppression which is “artisanal” and takes into account “individual minors’ needs,” and a one-size-fits-all policy.[56] However, the idea that there is not a middle ground here, such as creating more than one age category for minors to fit into, is not the strongest stance from which to argue. On a scale starting at no content-based protections for children whatsoever and ending with all of us uploading our faces and IDs every time we use the internet, surely there is a place where an acceptable balance is struck between accuracy, protection, and privacy.
VII. Conclusion
Despite this paper’s critiques of Goldman’s article, his argument overall is a necessary wake-up call at a time when moral panics around children’s safety online abound; when we can be overzealous to speak over the needs of actual children, and to trade in our own freedoms and those of our young people in the name of “protecting” kids. By limiting ourselves to segregate-and-suppress as a solution, we risk chipping away at what we may reasonably expect to remain private about ourselves. We risk censorship, culture war escalations, and a further narrowing of the publisher playing field. This critique is particularly pertinent as the Kids Online Safety Act, which has long been the subject of these very criticisms, works its way towards potentially being enacted into federal statute.[57] But Goldman’s article could have been further strengthened by taking more seriously the harms that do befall minors online, and by engaging more meaningfully with the breadth of legislation and policy which employs segregate-and-suppress tactics. Perhaps the most successful thing about Goldman’s paper is that it sounds the alarm on the risks inherent in our current path and encourages us not to view segregate-and-suppress as our best, last solution.
[1] Eric Goldman, The ‘Segregate-and-Suppress’ Approach to Regulating Child Safety Online, 28 Stan. Tech. L. Rev. 173 (2025), https://law.stanford.edu/wp-content/uploads/2025/07/Segregate-and-Suppress.pdf.
[2] Id. at 176-177.
[3] See Kids Online Safety Act, S. 1748, 119th Cong. (2025); see also Bipartisan Coalition of Attorneys General Urges Congress to Advance Senate Kids Online Safety Act, National Association of Attorneys General (Feb. 10, 2026), https://www.naag.org/press-releases/bipartisan-coalition-of-attorneys-general-urges-congress-to-advance-senate-kids-online-safety-act/.
[4] See Jackie Mader, Posted in Early Education, IPads in Kindergarten, YouTube Videos at Snack Time: Parents are Pushing Back on Screens in the Early Grades, The Hechinger Report (March 10, 2026), https://hechingerreport.org/ipads-in-kindergarten-youtube-videos-at-snack-time-parents-are-pushing-back-on-screen-time-in-the-early-grades/.
[5] Alex Vadukul, ‘Luddite’ Teens Don’t Want Your Likes, N.Y. Times (Dec. 15, 2022), https://www.nytimes.com/2022/12/15/style/teens-social-media.html; see also Alex Vadukul, Now in College, Luddite Teens Still Don’t Want Your Likes, N.Y. Times (Jan. 30, 2025), https://www.nytimes.com/2025/01/30/style/luddite-teens-reunion.html.
[6] See Anna Saavedra, Amie Rapaport, & Daniel Silver, Survey: Parents and Teens Support School Cellphone Bans, and Most Don’t Perceive Major Downsides, Brookings (Jan. 26, 2026), https://www.brookings.edu/articles/survey-parents-and-teens-support-school-cellphone-bans-and-most-dont-perceive-major-downsides/.
[7] See Andrew Arsht & Daniel Etcovich, The Human Cost of Online Content Moderation, Harv. J. of L. and Tech. (Mar. 2, 2018), https://jolt.law.harvard.edu/digest/the-human-cost-of-online-content-moderation.
[8] Id.
[9] See Katie Harbath, History of the Cambridge Analytica Controversy, Bipartisan Policy Center (Mar. 16, 2023), https://bipartisanpolicy.org/article/cambridge-analytica-controversy/.
[10] See Shoshana Zuboff, The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power (PublicAffairs 2019).
[11] See, e.g., Children’s Online Privacy Protection Act of 1998, 15 U.S.C. §§ 6501–6506.
[12] Goldman, supra note 1, at 222.
[13] See UK: New Molly Russell Documentary Provides Further Evidence that Social Media Needs Complete Redesign, Amnesty International (Mar. 5, 2026), https://www.amnesty.org/en/latest/news/2026/03/uk-new-molly-russell-documentary-provides-further-evidence-that-social-media-needs-complete-redesign/.
[14] See Jonathan Haidt, The Anxious Generation: How the Great Rewiring of Childhood Is Causing an Epidemic of Mental Illness (Penguin Press 2022).
[15] See, e.g., Clare Duffy, ‘There Are No Guardrails.’ This Mom Believes an AI Chatbot is Responsible For Her Son’s Suicide, CNN (Oct. 30, 2024), https://www.cnn.com/2024/10/30/tech/teen-suicide-character-ai-lawsuit; see also Kashmir Hill, A Teen Was Suicidal. ChatGPT Was the Friend He Confided In, N.Y. Times (Aug. 26, 2025), https://www.nytimes.com/2025/08/26/technology/chatgpt-openai-suicide.html; An Act to Regulate and Prevent Children’s Access to Artificial Intelligence Chatbots with Human-like Features and Social Artificial Intelligence Companions, L.D. 2162, 132nd Leg. (Me. 2026).
[16] See Amanda Hoover, Students Are Likely Writing Millions of Papers With AI, Wired (Apr. 9, 2024), https://www.wired.com/story/student-papers-generative-ai-turnitin/.
[17] See, e.g., Elliot Rodger: How Misogynist Killer Became ‘Incel Hero’, BBC (Apr. 25, 2018), https://www.bbc.com/news/world-us-canada-43892189.
[18] See Malu Cursino & Grace Dean, Social Media Plays ‘Huge Role’ in Promoting Traditional Gender Views, Say Experts, BBC (Mar. 5, 2026), https://www.bbc.com/news/articles/c4g55v1g95qo.
[19] See Brooke Tanner & Nicol Turner Lee, Children’s Online Safety Laws Are Failing LGBTQ+ Youth, Brookings (Jul. 9, 2025), https://www.brookings.edu/articles/childrens-online-safety-laws-are-failing-lgbtq-youth/.
[20] Goldman, supra note 1, at 176.
[21] Id. at 183-86.
[22] It is worth noting that this distinction between online and in-person spaces may be diminishing in some industries, as legalized marijuana businesses are increasingly required/incentivized to scan IDs rather than just viewing them, and at times, to store identifying data. See Why Do They Scan IDs at Dispensaries? ARCannabisClinic.com (Aug. 2, 2024), ttps://www.arcannabisclinic.com/post/why-do-they-scan-ids-at-dispensaries.
[23] Goldman, supra note 1, at 184.
[24] Id. at 191.
[25] Id.
[26] Id. at 193.
[27] Id. at 186.
[28] Id. at 187-188.
[29] Id. at 188-189.
[30] See id. at 195.
[31] Id.
[32] Id. at 209.
[33] See id. at 203.
[34] Id. “Bounce rate” is a term used to describe the rate in which users “bounce” off of, or give up on, accessing a website due to constructed barriers.
[35] Id. at 208.
[36] Id. at 217-219.
[37] Id. at 219.
[38] Id. at 222.
[39] Id. at 196.
[40] See, e.g., Jay Waagmeester, Lawmakers Take Another Try at Identifying What is ‘Harmful to Minors’, Florida Phoenix (Jan. 27, 2026), https://floridaphoenix.com/2026/01/27/lawmakers-take-another-try-at-identifying-what-is-harmful-to-minors/.
[41] See Goldman, supra note 1, at 201.
[42] Id. at 223-228.
[43] Id. at 230-231.
[44] See The Online Radicalization of Youth Remains a Growing Problem Worldwide, The Soufan Center (Sept. 9, 2025), https://thesoufancenter.org/intelbrief-2025-september-9/; Rita Chatterjee, Their teenage sons died by suicide. Now, they are sounding an alarm about AI chatbots, NPR (Sept. 19, 2025) https://www.npr.org/sections/shots-health-news/2025/09/19/nx-s1-5545749/ai-chatbots-safety-openai-meta-characterai-teens-suicide; Musafa Mohammed Hassan et al., The Association of Social media Use and Other Social Factors with Symptoms of Attention-Deficit/Hyperactivity Disorder in Egyptian University Students, 25 BMC Psychiatry 19 (2025).
[45] Goldman, supra note 1, at 228.
[46] Id. at 181.
[47] Goldman, supra note 1, at 176.
[48] Id. at 194.
[49] Id.
[50] Id. at 176.
[51] See Kaitlyn Huamani & Barbara Ortutay, Mark Zuckerberg Set to Testify in Watershed Trial Testing Social Media Addiction Claims, PBS (Feb. 18, 2026), https://www.pbs.org/newshour/nation/mark-zuckerberg-set-to-testify-in-watershed-trial-testing-social-media-addiction-claims; Bobby Allyn, Jury Finds Meta and Google Gegligent in Social Media Harms Trial, PBS (Mar. 25, 2026), https://www.npr.org/2026/03/25/nx-s1-5746125/meta-youtube-social-media-trial-verdict.
[52] Id. at 227.
[53] Id. at 226.
[54] Id. at 201.
[55] Id. at 228.
[56] Id. at 218.
[57] See Jenna Leventoff, ACLU and Students from Across the Country Urge Congress to Vote No on Kids Online Safety Act, ACLU (Jul. 24, 2024), https://www.aclu.org/press-releases/aclu-and-students-from-across-the-country-urge-congress-to-vote-no-on-kids-online-safety-act.