A river in a mountain valley covered in cash and coins instead of rubble, with a mudslide pouring down from the mountains in the background

DESCRIPTION: Meta is paying up to $17 billion, thereby bringing the lawsuit over harm to young users to an end after eight days. Why time limits and night-time restrictions are the wrong way to tackle the problem.

A trial that lasted just eight days

On 18 August 2026, the trial brought by 51 US state attorneys-general against Meta began in Oakland. By 25 August, it was over. The company will pay up to $ 17 billion over 10 years and implement a series of safeguards for minors on Instagram and Facebook. No judgement was handed down.

A settlement brings the proceedings to an end without resolving the matter.

A settlement clarifies nothing. It ends the dispute whilst leaving the contentious issue unresolved, as does the question of guilt. The party paying admits nothing, and the party accepting the settlement makes no findings. In this case, this happened in the middle of the evidence-taking phase: the case had been running since 2023, the jury had been sitting in the courtroom of the US District Court for the Northern District of California for a week, and the opening statements had been delivered. Then the parties reached an agreement.

The allegation was that Meta had designed and rolled out features that drive children and young people to compulsive use, and had misled users, families and the public about the extent and severity of these risks. Added to this was the allegation that, for years, it had collected data from children under the age of thirteen without parental consent. Legally, the lawsuit was based, among other things, on the Children’s Online Privacy Protection Act and California competition and advertising law.

As a corporation with sufficient funds, Meta is thus buying its way out of the evidence-gathering process and the determination of liability. This is done legally and in the prescribed manner. The settlement is the instrument. Anyone who can raise a sufficient sum ensures that no further witnesses are heard, no internal documents are read out, and no responsibilities are assigned. The ability to bring proceedings to an end before any findings are made depends on one’s financial resources.

What Meta has agreed to in detail

The list is long and, in parts, unusually specific. For users under the age of eighteen, a daily time limit of two hours will apply in the future, and only a parent can lift it. Between midnight and 6 am, the apps will be blocked and can only be unlocked by a parent. Notifications will be blocked at night from 10 pm to 7 am and during school hours from 8 am to 3 pm. Like and reaction counts will no longer be displayed to minors. Filters that simulate cosmetic procedures are prohibited for this group. Meta must introduce age verification, remove children under the age of thirteen from its platforms, respond to ninety per cent of reports of harmful content within six hours, and allow an independent auditor to access information, report regularly and contact the public prosecutor’s offices directly. A court order prohibits the company from making false or misleading statements about its own safety features in future.

Two clauses deserve particular attention. The time limit is reduced to one hour, and the night-time ban is extended from 10 pm to 7 am, but only if other platforms make comparable commitments. The protection of children from harmful users therefore depends on the behaviour of its competitors. Furthermore, minors are given the option to choose a feed without active algorithmic personalisation. The algorithm remains the default setting, and behavioural research has revealed enough about its impact.

The knowledge is there, but accountability is lacking.

The usual objection to settlements is that they keep information under wraps. In this case, however, that objection does not apply. Internal studies on the body image of teenage girls have been in the public domain since the Haugen documents were released, as have unsealed passages from the statements of claim, and Mark Zuckerberg gave evidence before a jury in February 2026. The material is accessible.

What is missing, however, is attribution. No court has determined who made which decision and when, which warning was ignored, and who was aware of which figures. No precedent is being set that the next case could draw upon. A jury in Los Angeles had ruled differently in March 2026, holding Meta and Google responsible for a young woman’s anxiety and depression and awarding her six million dollars in damages. A trial can, therefore, conclude. In Oakland, this did not happen.

What the addiction model fails to explain

The widespread interpretation that social media harms young people due to excessive use is itself a form of defence mechanism. It transforms an economically engineered relationship into a question of self-control. The child is portrayed as a high-risk user who scrolls excessively and stays online late.

The addiction model explains why someone keeps coming back. It does not explain why coming back is agonising. To speak of addiction implies a substance that has an effect, a dose that increases, and a withdrawal that eventually ends. The experience at issue here is not intoxication. It is tension: having to check, being scrutinised, not measuring up, checking again. That is the hallmark of a superego.

The platform as a superego machine

The superego does not merely say ‘you must not’. It demands: ‘Show yourself, react, be visible, remain accessible, do not miss a thing.’ Therein lies its relentlessness. One can obey a prohibition and then find peace. A demand for visibility can only be met, ever further.

Jacques Lacan summarised this trait in Seminar XX: no one forces anyone to enjoy themselves except the superego, and its imperative is ‘Enjoy!’. In doing so, the superego sheds its role as a mere prohibitive authority. It drives us on. Anyone who follows the demand never reaches a point where they have enjoyed enough, for the demand specifies no measure. It repeats itself endlessly.

Instagram and Facebook are superego machines. The classical superego is internalised, and because it lies within, it is negotiable, shiftable, open to interpretation, and, in the best-case scenario, can be appeased. This is what all therapeutic work thrives on. The platform re-establishes the same authority outside the subject, quantifying it and making it public. You cannot argue with a number. An accusation could be refuted. A figure stands there.

Every metric can be surpassed.

A number can always be higher, a reaction faster, an image better. The ego ideal used to remain vague and, precisely for that reason, allowed room for manoeuvre. Here, it becomes precise yet remains unattainable, because no state is defined that could be considered sufficient. There is no saturation point. At best, there is a reprieve.

Lacan coined the term ‘object a’ for that which desire is directed towards: a cause that vanishes the moment it is attained. You get what you wanted, and yet it wasn’t it after all. Metrics illustrate this structure with particular clarity because they also count the absence. Added to this is his second thesis: that human desire is the desire for the Other. What is desired is to be desired by the Other. It is precisely this desire that is reflected in the numbers of ‘likes’.

For Lacan, the Big Other is the symbolic order that stands for meaning, and for him it never stands for it without gaps: there is no Other of the Other. A feed, on the other hand, appears without gaps. It has a number for every question. An Other without gaps does not provide any additional stability. Instead, it takes away from the subject the place where it should have established itself.

The ‘like’ count goes through Baudrillard’s four stages.

In 1981, Jean Baudrillard described four successive stages of the image. First, the sign mirrors a reality. Then it conceals and distorts it. Then it conceals the fact that there is none. Finally, it no longer has anything to do with reality and is its own simulacrum. His favourite example was money, from gold through banknotes backed by gold to electronic transactions backed by nothing at all.

The number beneath an image follows the same path. At the start, it counts the approval of people who have given their approval. Then it takes their place and counts things that are not approval: outrage, chance, sharing. Then it mainly indicates how often the algorithm has displayed the image and says nothing more about appreciation, since there is no one behind it to express such appreciation. Ultimately, it generates what it purports to measure. Those with many ‘likes’ receive encouragement for the numbers. They are bought, farmed, optimised.

This casts the concealment of ‘like’ counts in a different light. Meta removes from view a symbol that has long since ceased to stand for anything, and lets the algorithm do its work, which generates it. Reach remains palpable; the comparison takes place. It simply becomes unverifiable.

The figure of 17 billion can also be interpreted in this way. A sum for damages that no court has awarded refers to nothing specific. It is the amount that two parties have agreed upon, with a figure on the table. The number represents a responsibility whilst leaving it undefined.

Read receipts, streaks and viewer lists.

This is not merely a metaphor. It is spelt out in user interfaces. The ‘read’ confirmation transforms silence into an action: previously, someone might not have replied; afterwards, they have not replied. The streak, which expires upon interruption, turns attention into a debt to be repaid by a set deadline. A story’s viewer list shows by name who has seen it and failed to react, thereby archiving rejection. In Seminar XI, Lacan describes the gaze as something that can come from anywhere and has no place where it can be pinned down. The viewer list gives it a place and a list of names. What would otherwise remain diffuse observation becomes a list of those who have seen it and remain silent.

None of these functions is accidental. Each technically implements a psychic operation, and each measurably intensifies the return. The superego machine consists of these individual parts. The court-mediated settlement does not touch upon any of them.

In adolescence, the gaze of peers matters

For young people, this carries greater weight than for adults. Not because they are more psychologically vulnerable, but because, at this stage of life, body image, desire, separation and belonging are renegotiated – specifically through the gaze of their peers. Separation from parents requires a reference point against which the self-image can be reshaped. The platform offers them a hierarchy.

Lacan’s essay on the mirror stage describes how the ego forms itself in relation to an external image that appears more complete than the self’s experience. The image precedes the subject, who follows it. A profile works with the same material and adds something: the image returns with an evaluation. Anyone who forms their identity there does so in relation to an image whose value has already been determined.

Added to this is the question of reversibility. A row in the school playground ends when those involved go home. What happens on a platform remains accessible, quantifiable and visible to third parties, and the circle of observers is indefinite. Anyone who presents themselves falsely at this stage loses the opportunity to leave it at that. This is precisely what someone needs when they are exploring who they might be.

The comparison leaves the question of causes open.

At this point, it would be disingenuous to present the matter as settled. Jonathan Haidt attributes rising rates of anxiety and depression to smartphones and social media. Candice Odgers, Amy Orben and Andrew Przybylski counter that the correlations in the data are small, mixed or non-existent; a much-cited study by Orben and Przybylski puts the contribution of digital technology to differences in well-being at no more than 0.4 per cent. Haidt and his colleagues, in turn, dispute the validity of these studies. The scientific debate is still ongoing.

That is precisely why the settlement is a loss. Proceedings involving a full taking of evidence would have compelled access to precisely the data that researchers lack: internal impact assessments, A/B tests and cohort analyses, which Meta possesses but is now refusing to disclose. The unresolved question is merely quantified, not answered. If one does not know the extent of the damage, it is all the easier to assign some arbitrary figure to it.

The restrictions tackle the problem in the wrong way.

Time limits, night-time curfews, throttled notifications: these are prohibitions. They appeal to the old form of the superego, to the ‘Thou shalt not’. To use Lacan’s terms, the settlement combats the prohibitive instance whilst allowing the driving force to continue unchecked. The platform’s imperative is ‘Show yourself, don’t miss a thing’, and a night-time ban does not affect that. It does not abolish it; it merely postpones it. At the same time, it creates the very thing on which it thrives: a period of time during which something happens that one has not seen. Anyone who is only allowed back in at seven in the morning has a whole night to catch up on.

Even the option for parents to override the restriction offers only questionable protection. The dispute over boundaries shifts into the family sphere, of all places, at precisely the stage of development when parental control is least effective and breaking free from it is a central developmental issue. The company refuses to take on the role of the responsible party and instead places the burden on the family’s dinner table.

What could really help

A night without notifications is better than a night with them. A truly independent auditor with access to documents and a direct line to the public prosecutors’ offices represents real progress, because it enables ongoing oversight. However, they check compliance with the agreed points rather than whether those points are the right ones.

An experience that predates this issue speaks against this form of regulation. Environmental law has, for decades, been familiar with everything that this agreement introduces: limit values, measurement requirements, reporting obligations, independent auditors, enforceable conditions and substantial fines. For fracking, there are methane regulations and wastewater regulations; for data centres, energy consumption is documented, whilst in the US there is a lack of binding efficiency standards and uniform regulations on water reuse. This regulatory framework has not prevented any of the damage in question. It has broken it down into authorised, documented and costed individual processes. A compliance regulation governs the process of data utilisation. Still, it has no scope, and there is no reason to assume that it works any differently on Instagram than it does for a borehole.

A genuine solution would be to block personalised advertising aimed at minors. A non-personalised feed should be the default setting for all minors. The verifiability of recommendation systems would complement this by conducting independent research and banning all features that log rejections, impose time limits on engagement, or make silence visible. In other words, regulation of the system’s design rather than the duration of its use.

Until that happens, the settlement remains what it is in form: a payment that sorts out many things but establishes nothing. The 17 billion is the price for not being condemned, and Meta can afford to pay it. It may alleviate individual suffering whilst simultaneously confirming that suffering must first arise, be measured, priced and compensated before it counts politically.

Key points in brief

• The trial in Oakland began on 18 August 2026 and ended on 25 August with a settlement. 51 state attorneys-general were involved. The payment amounts to up to $ 17 billion over 10 years and is subject to court approval.

• Commitments include, amongst other things, a two-hour daily time limit for minors, a night-time block from midnight to 6 am, notification blocks at night and during school hours, hidden ‘like’ counts, a ban on cosmetic filters, age verification and an independent auditor.

• The feed without algorithmic personalisation is an option. The algorithm remains the default setting.

• The time limit and night-time ban will only be tightened if other platforms follow suit. Protection depends on the behaviour of the competition.

• No judgement is handed down. The evidence regarding Meta’s knowledge is largely public; there is no judicial attribution of liability, and no precedent is set.

• Those with sufficient funds can buy their way out of the taking of evidence and the question of guilt. A settlement is the intended instrument for this, and the ability to terminate proceedings before any findings are made depends on one’s wealth.

• The addiction model explains the recurrence but leaves the tension unexplained. As a superego machine, the platform demands that it be visible from the outside, and the metric of this demand is open to being outdone, and is therefore never fulfilled.

• According to Baudrillard, the ‘like’ count has completed its journey from image to simulacrum. It now generates what it purports to measure. Its concealment removes a sign that no longer stands for anything, whilst leaving the sorting mechanism untouched.

• Lacan defines the superego as a driving force with the imperative ‘Enjoy!’. This force knows no bounds, and the platform caters precisely to it.

• Time limits and bans address prohibitions. The imperative of visibility remains untouched, and the night-time ban additionally creates a sense of what has been missed, which must be made up for in the morning.

• Verifiable, enforceable and sanction-backed requirements have existed in environmental law for decades. They have not prevented the damage; they have broken it down into authorised individual processes. Compliance regulates the process of utilisation, not its scope.

• An effective regulation would have to address the platform's design: default settings, advertising, the verifiability of recommendation systems, and a ban on features that make rejection visible.

Sources

· Attorney General Bonta Secures Transformative $17 Billion Settlement with Meta — Press release from the California Attorney General’s Office, 25 August 2026, containing the full list of conditions

· Proposed Settlement / Consent Judgment (PDF) — the full text of the settlement

· Attorney General Jay Jones Announces a $17 Billion Landmark Settlement with Meta — Analysis from Virginia

· Meta Reaches $17 Billion Settlement With States in Landmark Trial Over Teen Social Media Addiction — coverage of the conclusion of the proceedings

· Meta is back in the courtroom to face its biggest social media addiction trial yet — CNN on the start of the trial on 18 August 2026

· 29 States Take Meta to Trial: First Federal Social-Media MDL Case Reaches a Jury — Status of the proceedings and background

· US jury finds Meta and Google liable in social media addiction trial — the jury verdict from Los Angeles, March 2026

· Is There a Connection Between Teens’ Social Media Use and Their Mental Health? — An overview of the Haidt/Odgers controversy

· Why Three Studies by Vuorre and Przybylski Should Not Be Used to Assess the Impact of Social Media on Adolescent Mental Health — Haidt and colleagues versus the ‘all-clear’ studies

· Data Centres: Energy Use and Efficiency (CRS R49057) — Congressional Research Service on energy requirements

· Data Centres and Water Use (CRS R48646) — Congressional Research Service on water consumption

· Jacques Lacan: The Seminar, Book XX: Encore (1972/73) — the superego as the imperative of jouissance

· Jacques Lacan: The Seminar, Book XI: The Four Fundamental Concepts of Psychoanalysis (1964) — the gaze as object, the desire for the Other

· Jacques Lacan: The Mirror Stage as the Formative Principle of the Ego-Function (1949), in: Écrits I

· Jean Baudrillard: Simulacra and Simulation (1981) — the four stages of the image and the precession of simulacra