The Origins and Consequences of Big Tech’s “Tobacco Moment”  

The constitutional debate over social media bans seems to be moving away from regulating speech and content toward scrutinizing product designs and their resulting harms. Once bans are viewed through that lens, federal and state legislatures in the United States take on a much broader constitutional authority to protect children in their “traditional social environment” online than many First Amendment debates would suggest. 

From Content to Design: The Changing Landscape of Legal and Theoretical Arguments  

In March, a California jury found Meta and Google jointly liable (with a 70–30 split) in a social media addiction trial awarding six million dollars to the plaintiff, twenty-year-old K. G. M. Her testimony suggested that her use of YouTube and Instagram since ages six and nine respectively (sometimes, for up to sixteen hours a day) led to anxiety, depression, body dysmorphia, and other disorders.   

But this is just one of the many cases currently unfolding in courts. Such an approach has prompted American legislators to argue that addictive platform design justifies age-based restrictions similar to those for other products that pose significant public-health risks. Current federal initiatives like KOSMA and the Stop the Scroll Act serve this purpose, as well as the many state proposals and laws already adopted, e.g., in Florida, California, Illinois, or Rhode Island, to regulate children’s use of these platforms. 

The Ethics and Public Policy Center’s Clare Morell aptly wrote that recent big tech cases have shifted from content to product design features, and consequently, big tech is facing its “tobacco moment” born out of the following question (as Morell put it): “Did these companies negligently design and market a highly addictive product to children and did they know—and fail to warn users—that their products cause harms to minors?” 

I would argue that the better question is whether the companies themselves create a harmful and unhealthy social environment for children. 

Other nations have attempted to answer this question. Most notably, French legislators adopted a nationwide social media ban for kids a few weeks ago, becoming the first European country to do so; they set the age of “digital majority” for children at fifteen years. The law argues that children as a protected class are too young to access social media networks. It provides a broad, platform-neutral scope, framing the introduction of digital majority as a public health necessity to aid children’s civic development. The controversial nature of any such legislation is demonstrated by the French Constitutional Council’s August 14th decision to declare the law unconstitutional by, among other things, introducing unnecessary age verification requirements for people over the age of digital majority without necessary data privacy safeguards.   

The French public health narrative is reminiscent of Hungarian law. At the turn of the millennium, the Hungarian Constitutional Court introduced a prohibition on tobacco ads in national media, holding that that the right to health, just as the right to a healthy environment, serves the wellbeing of our future generations. The court argued that public health considerations justify prohibitions that promote the welfare of the protected population. In addition, in a 2021 immigration case, the same Court presented the idea that in protecting human dignity, the state needs to protect the “traditional social environment of individuals” as one that “defines the development of one’s personality, the direction and frames of their identity and as such needs to be considered within the bounds of the quality of human life.” (I shall use the adjective “traditional” to refer to the ordinary interpersonal environment in which people grow and develop.) 

Against this background, an assessment of the legal, policy, and theoretical landscape surrounding this topic in the United States is timely. Applying the two (French and Hungarian) approaches to shape present American debates on product liability and legislative bans is useful for two reasons. First, it may provide both federal and state legislatures with additional reasons to enable or exercise police powers supported by common-good considerations. (The U.S. constitutional law doctrine of police powers refers to the states’ general legislative authority to regulate health, safety, welfare, and public order—powers the federal government does not possess in the same general form.) Second, they point to the necessity of enacting effective consumer and child protection regimes that shield the “traditional social environment” of children online, rolling back some attention-driven and engagement-focused platform features that are addictive and that endanger younger generations. 

In other words, the current modus operandi of social media is harmful to the development of children’s “traditional social environment” online, which was healthy before creative destruction brought about unintended consequences. 

A possible counterargument could be that it is hard to define the “traditional” social environment online, especially in the social media context. But sociological literature has already described the evolution of important differences in kind, e.g., from traditional, asynchronous methods of mass communication to synchronous methods, resulting in the increasing prevalence of social media in recent years. 

The concept of a “traditional social media environment” should be understood relatively rather than absolutely. I do not mean an idealized or nostalgic vision of social life online. Rather, “traditional” means the prevailing communicative ecology in which children’s personality, identity, and social relationships develop at a given historical moment (as a baseline). Scholars like Manuel Castells in The Rise of the Network Society have long described successive transitions—from one-way mass media to participatory social media. As communication technologies have evolved over time—from broadcast media to networked social media and now to AI-mediated interaction—that baseline itself changes. My argument is that generative AI represents a qualitative break in that evolution, making the pre-AI platform ecosystem the relevant “traditional” baseline for contemporary constitutional analysis. 

A great example is the harm AI influencers cause to the healthy development of one’s personality and identity, and to the “traditional social environment” itself. There is evidence that their emergence affects children’s socialization, mental health, and identity formation, distorting their traditional social environment online. Because if children’s socialization happens primarily online, then how “traditional” can it really be? If they interact more and more with AI entities that express feelings and personalities and communicate opinions and yet are not human but can only be distinguished from actual humans with great difficulty even by an adult, what is this doing to children’s social development? 

In terms of the current modus operandi of the platforms, we also need to look at children as captive audiences. By product design, they are the main targets in terms of advertising and potential future purchasing power, and they are uniquely vulnerable. Unlike newspapers, television, or even the early internet, algorithmic platforms continually adapt to the user’s behaviors and preferences, reducing the practical significance of the user’s theoretical freedom to leave. 

AI-driven advertising captivates users through algorithmically tailored content; it feeds viewers content based on their former searches (i.e., their taste profile). Children may be particularly vulnerable to exposure to harmful content, as evidenced by cases like Anderson v. TikTok, which resulted from the tragic death of ten-year-old Nyla Anderson due to an AI-recommended “blackout challenge.” Her parents sued TikTok for, among other things, negligence and products liability, alleging that TikTok knowingly allowed such videos and affirmatively recommended them to minors through its algorithm.

As this case suggests, AI has unforeseen capabilities in learning and adapting to user preferences, which can reinforce engagement with specific—often unlawful and harmful—content. Thus, the central issue in litigation is no longer merely content, but the design choices that amplify exposure to it. 

When advertising is directed toward specific audiences looking at specific desired content, non-skippable video ads appear for specific groups of users. In other words, while trying to access content relevant to their filter bubbles, these groups become captive audiences of such AI-driven, recommended advertising. Once the platform becomes aware of their preferences, children can’t escape content directly marketed to them under their taste profiles. Even though they might be able to delete the app, they most often will not, as they are driven by a need to participate in the discourse (driven by the fear of missing out—“FOMO”—and the network effect) in their respective groups. They become a captive audience, because if they leave, they miss out on information they think might be vital to their socialization. There is nothing “traditional” about this environment. 

To many children, social media platforms function as their traditional social environment, as a place where they can express themselves. Children should have, in the online world, the protections afforded them in physical public spaces. There should be certain “places” where children can avoid offensive speech; but if there cannot be any, then prohibitive rationales will emerge, arguing for bans on social media to enable children to avoid this problem altogether. Different child-protective and public health or welfare rationales support these arguments, elaborated—in the context of the First Amendment—in Prince v. Massachusetts, which held that “It is the interest of youth itself, and of the whole community, that children be both safeguarded from abuses and given opportunities for growth.”  

From Scanning to Banning: The Changing Landscape of Policy and Legislation 

Debates about content moderation (under Section 230 of the Communications Decency Act) persist in the United States, but the “captive audience” angle remains unexplored. As someone who was educated under a different free-speech doctrine in Europe, I find this inquiry worthwhile because creative destruction also requires the adaptation of established mechanisms (as well as doctrines) to new contexts. Also, this doctrine may serve as a foundational rationale when framing social media bans for kids.  

It is now up to the American courts and legislators to create solutions that balance constitutional liberties with the need to protect children from preventable tragedies. The first judicial and legislative solutions are just now being tried in the court of public opinion and in actual courts as well. 

Whether courts will ultimately uphold social media restrictions remains uncertain. What is already becoming clear, however, is that the constitutional conversation is changing. As litigation increasingly treats social media platforms as products rather than merely speakers, debates once dominated by content moderation may increasingly turn on product design, public health, and the state’s traditional responsibility to protect children. If that transition continues, Big Tech’s “tobacco moment” will prove to be more than a metaphor—it will mark a constitutional realignment, as justifications for regulating social media seem to migrate from the First Amendment toward police powers, product liability, and public health laws. 

Image licensed via Adobe Stock.