For a long time, the case against social media was argued far from the courtroom. It belonged to parents, psychologists, teachers, researchers and, sometimes, users themselves. Platforms were accused of fragmenting attention, invading the night, turning social comparison into a permanent reflex and making one apparently simple gesture strangely difficult: putting the phone down.

That debate has now entered another dimension.

In the United States, Meta is facing a legal offensive whose implications extend far beyond Facebook and Instagram. On August 18, 2026, a federal trial opened in Oakland involving the company and a coalition of U.S. attorneys general. Twenty-nine states are participating in the action, accusing Meta, among other things, of designing aspects of its services to prolong engagement among minors while downplaying or concealing associated risks. Meta disputes those allegations and points to the numerous measures it has introduced to protect teenagers.

Behind the legal battle, however, lies a much larger question.

What happens when an economy built on attention discovers that it has become a little too good at holding it?

From engagement to dependence

Social networks did not invent the human desire to be seen, recognized or approved. Nor did they create curiosity, distraction or social comparison. Their fundamental innovation lies elsewhere: they transformed those behaviors into measurable data, and then turned that data into industrial systems of optimization.

Every click produces a signal. Every pause in front of a video tells the system something. Every share, comment, search, replay or abandonment helps refine its understanding of what keeps the user engaged.

As platforms learn, the experience therefore ceases to be the same for everyone. It becomes personal.

The feed has virtually no end. One video leads to another. A notification brings the user back into the application. Recommended content evolves according to previous reactions. The product continuously observes what works and adjusts what it offers.

Taken individually, none of these mechanisms is sufficient to establish addiction. Together, however, they create something unprecedented at this scale: a digital environment capable of continuously learning how to retain the attention of billions of individuals.

This is precisely where the debate changes.

The user is no longer simply facing a product they choose to consume. They are interacting with a system that gradually learns to anticipate their preferences.

The asymmetry is considerable.

The business model behind the screen

To understand the problem, one must return to a much simpler reality: attention has economic value.

Most major social networks offer their services to users free of charge because their businesses are financed primarily through advertising. The better a platform understands its audience, the more precisely it can offer advertisers targeting and measurement capabilities. And the longer users remain present and the more frequently they return, the more opportunities exist to place commercial content in front of them.

This does not mean that every additional minute spent on Instagram mechanically becomes revenue. But it creates a structural convergence between engagement and economic performance.

The consequence is fundamental.

For years, the technology industry treated “engagement” as an almost inherently positive metric. A user who returns frequently appears satisfied. Content watched for longer appears relevant. A feature used intensively appears successful.

But what happens when engagement is no longer entirely voluntary?

The boundary between an excellent product and a compulsive one becomes much harder to define.

That is the boundary American courts are beginning to explore.

The trial of design

The proceedings against Meta matter because they are gradually shifting the discussion away from content and toward the architecture of the platforms themselves.

The federal trial now underway in California concerns, among other issues, allegations that Meta misled the public about the safety of its services and violated U.S. rules governing children's data. The states are also seeking changes that could significantly affect how the platforms operate. Meta rejects the claim that it deliberately sought to make its products dangerous or addictive and disputes the interpretation placed on some of its internal documents.

The legal distinction is essential: allegations are not facts established by the outcome of a trial.

Yet the very fact that courts are examining them opens a considerable breach.

If a platform can potentially be held responsible not only for what a user publishes but also for the way the product itself is designed, part of the legal architecture that accompanied the rise of social media could be reconsidered.

The debate would no longer concern moderation alone.

It would concern the product.

Is infinite scroll neutral? Is autoplay? At what point does a notification cease merely to inform and begin to solicit? How far can a recommendation system optimize the retention of a teenager before that optimization becomes a safety issue?

Questions that once belonged primarily to the ethics of design could increasingly become questions of law.

What Meta knew

The issue becomes even more sensitive when it intersects with research conducted inside the companies themselves.

For several years, documents originating from Meta have fueled debate over what the company knew about the effects of its platforms, particularly on younger users. By early 2026, a research project at NYU Stern School of Business had catalogued information concerning 35 internal studies attributed to Meta examining links between its products and various risks to young people's mental health. The material includes documents disclosed by whistleblowers as well as evidence emerging through litigation.

Their interpretation remains disputed.

A technology company can legitimately investigate the potential negative effects of its own products precisely because it wants to mitigate them. The existence of internal research is therefore not, by itself, evidence of an intention to cause harm.

The harder question is what happens next.

When a company identifies a risk, what does it do about it?

That is where much of the battle lies.

Plaintiffs are seeking to demonstrate that certain risks were known while mechanisms encouraging engagement continued to be deployed. Meta argues, by contrast, that documents have sometimes been presented without adequate context and points to measures introduced over the years to improve safety for teenagers.

The courts will have to assess those competing narratives.

Society, meanwhile, has already begun a trial of its own.

The tobacco precedent

The comparison with the tobacco industry is almost inevitable.

It must also be handled carefully.

Instagram is not a cigarette. The biological mechanisms, medical risks and patterns of use are fundamentally different. Social networks can also generate genuine utility: maintaining relationships, discovering knowledge, developing a professional activity, accessing communities or simply providing entertainment.

The analogy becomes more interesting, however, when the historical trajectory of the two controversies is considered.

For decades, responsibility for smoking was framed largely as an individual matter. Consumers chose to smoke. Then scientific evidence accumulated. Internal corporate documents became important. Marketing practices came under scrutiny. The question gradually shifted from the behavior of the consumer toward that of the producer.

A comparable dynamic could emerge around digital platforms.

“Just close the app” appears to be a sufficient answer as long as the product is considered passive.

It becomes less convincing when the product is specifically engineered to make leaving less likely.

The particular problem of children

The question becomes harder still when minors are involved.

An adult can theoretically understand that a platform is competing for their attention. They can disable notifications, delete an application or impose limits on their own use.

Expecting the same degree of self-control from a twelve- or thirteen-year-old is considerably more controversial.

This is one reason the current legal battle focuses so heavily on younger users. Attorneys general have accused Meta of attracting and retaining children on its platforms while failing to comply adequately with certain requirements applying to users under thirteen. Even before the August trial opened, U.S. District Judge Yvonne Gonzalez Rogers had declined to dismiss the case and sided with the states on part of their argument concerning parental-consent requirements under the Children's Online Privacy Protection Act.

And Meta is no longer facing a handful of isolated lawsuits.

Families, public authorities and school systems have brought cases against several major platforms. More than a thousand U.S. school districts have participated in litigation arguing that they are bearing part of the consequences of the youth mental-health crisis and seeking, among other remedies, changes to features such as infinite scrolling and autoplay.

The conflict is therefore beginning to move beyond the relationship between a company and its users.

It is becoming a question of externalities.

Who pays when a free product generates costs elsewhere that are anything but free?

An entire industry in the dock

Focusing exclusively on Meta would nevertheless be a mistake.

Instagram and Facebook occupy a central position in the current proceedings, but the attention-economy model extends far beyond a single company. TikTok, Snapchat and YouTube have also faced litigation concerning their alleged effects on younger users.

That is what makes the present moment particularly significant.

If courts ultimately establish that a platform can be held liable for design characteristics that encourage compulsive behavior, the precedent is unlikely to remain confined to Meta.

The entire industry would have to look at its interfaces differently.

A seemingly innocuous product decision could acquire legal significance. Teams tasked with maximizing retention would have to work under greater constraints from those responsible for safety. Time-spent metrics could become less unquestionably desirable. Products designed for minors could evolve toward architectures fundamentally different from those offered to adults.

Most importantly, regulation could begin to intervene not merely in prohibited content but in the mechanisms organizing its consumption.

That would represent a paradigm shift.

Can attention be regulated?

This is precisely where the difficulty begins.

Banning a dangerous substance is relatively easy to conceptualize. Regulating a universal psychological mechanism is much harder.

How many consecutive videos are too many?

How many notifications?

At what hour should a platform stop trying to bring a teenager back?

Should infinite scrolling be removed? Should interfaces deliberately introduce friction? Should mandatory pauses be imposed? Should recommendation algorithms be less personalized for minors?

And who determines the acceptable level?

The state?

Parents?

Companies?

The user?

Every answer creates problems of its own. Regulation that is too weak risks becoming cosmetic. Regulation that is too intrusive could infantilize adults, restrict legitimate uses or give governments considerable influence over the architecture of digital spaces.

There is probably no single mechanism capable of solving the problem of digital addiction.

That does not mean the current architecture is the only possible one.

The real revolution may be economic

Behind the lawsuits lies an even more uncomfortable question.

What if the problem is not merely the design of social networks, but their economic model?

As long as the value of a platform depends significantly on its ability to capture, measure and monetize attention, the company retains a natural incentive to increase engagement.

It can make that engagement safer.

It can protect minors more effectively.

It can introduce limits.

But it cannot easily abandon the principle of retention without touching one of the central mechanics of its business.

This may be the fundamental contradiction of the attention economy.

We are asking platforms to help us use them less while part of their economic power rests precisely on our willingness to use them more.

A trial without appeal?

The judicial verdict remains unwritten.

Meta has arguments of its own, vigorously contests the allegations and points to the investments it has made in safety and tools for teenagers. The relationship between social media and mental health also remains complex: correlation, causation, individual vulnerabilities, family environments and the nature of specific patterns of use make universal conclusions difficult.

It would therefore be premature to pronounce a definitive scientific or judicial judgment.

But the historical trial of the attention economy may already have begun.

For two decades, the success of digital platforms was measured by their ability to attract users, bring them back and extend the time they remained.

We are now discovering the other side of that performance.

A technology sufficiently effective at retaining our attention eventually raises a question the digital industry would probably have preferred to leave to each of us:

if leaving becomes difficult, are we still simply choosing to stay?

Main Sources

Reuters — Meta rejects claims it sought to hook children to Facebook, Instagram, August 18, 2026.

California Department of Justice — proceedings brought by state attorneys general against Meta and pretrial ruling, June 30, 2026.

Associated Press — overview of major proceedings involving Meta and other platforms concerning alleged harms to young users, August 18, 2026.

The Wall Street Journal — opening of the federal trial concerning child safety and Meta's practices, August 18, 2026.

NYU Stern Tech and Society Lab — Meta's Internal Research, compilation of publicly disclosed internal research, updated March 2026.