On 18 August, opening statements begin in the federal court in Oakland in a trial against Meta that could affect Facebook and Instagram far more deeply than another large fine.
Twenty-nine US states accuse the company of illegally collecting and using the data of children under the age of 13, while California, Colorado, Kentucky and New Jersey are also bringing claims in court that Facebook and Instagram were designed to encourage compulsive use by young people and that Meta publicly minimised the risks of such design.
CEO Mark Zuckerberg and Instagram head Adam Mosseri are expected to be among the witnesses. The case is being presided over by federal judge Yvonne Gonzalez Rogers, who will make the final decision with the assistance of an advisory jury selected last week.
The states are seeking far more than a financial penalty. Proposed measures include age restrictions, time limits for young users, the removal of infinite scroll and certain notifications, changes to how content is recommended, and even the deletion of algorithms and AI models created using children’s data.
The court has yet to grant any of these requests, but the fact that they have reached a federal judge shows how far the legal debate over social media has progressed.
Before the court, the issue is no longer only what appeared on the platform and who published certain content; decisions made by the company itself while building the product have become the subject of the proceedings.
When design becomes a matter of responsibility
For almost three decades, American technology companies have operated under the protection of Section 230 of the Communications Decency Act. This law effectively prevents an internet platform from being treated as the publisher or author of content posted by its users.
For Facebook, YouTube, X, Reddit and thousands of other services, that protection has been one of the most important legal foundations for operating in the United States.
Its limits, however, are far more complex than the popular notion that platforms are not responsible for content on the internet. On 28 April this year, the Ninth Circuit Court of Appeals upheld a broad interpretation of Section 230 in a case against Meta concerning violence against the Rohingya in Myanmar.
The selection, ranking, and recommendation of third-party content can constitute a publishing function protected by Section 230
The court held that the selection, ranking, and recommendation of third-party content can constitute a publishing function protected by Section 230, even when that selection is performed algorithmically.
In separate opinions, two judges openly expressed their displeasure at the extent to which case law has expanded the original scope of that protection.
That limits the ability to simply declare any algorithm a defective product and thereby circumvent federal protections. Courts must determine which obligation the plaintiff seeks to impose on the company.
If the claim is essentially that Meta should have selected, ranked or removed its users' content differently, Section 230 remains a very strong defence. Infinite scroll, night-time notifications, reward systems based on likes, autoplay of content, methods of age determination, and procedures that prolong the time minors spend in the application open up a different legal avenue.
Here, the company's behaviour is easier to distinguish from the specific post or video created by the user, and it is at this boundary that much of the next generation of litigation against social networks will arise.
States are attacking what Meta has created
The coalition of state prosecutors claimed in 2023 that Meta was adapting its business model to maximise the time and attention of young users. In their lawsuit, recommendation systems, constant notifications, social comparison through likes, visual filters and infinite scroll are cited as problematic elements.
The states also claim that Meta publicly presented the platforms as safer for young people than, based on their interpretation of internal documentation and product behaviour, the company had reason to do. Meta disputes these allegations and maintains that it has been developing protections for minors for years.
In late June, Judge Gonzalez Rogers rejected Meta's attempt to strike key claims before trial. She concluded that there are factual issues for the court to resolve, including whether the platforms foster compulsive use and whether certain company statements about their design and security may have misled consumers.
COPPA regulates the collection of data from children under 13, while deceptive business practices concern the company's own statements and actions
The court also accepted the states' argument that the way Meta obtained parental consent was insufficient to meet the requirements of the federal Children's Online Privacy Protection Act, known as COPPA.
That part of the process is particularly dangerous for the company because it does not rely on user-generated content. COPPA regulates the collection of data from children under 13, while deceptive business practices concern the company's own statements and actions.
Section 230 provides far less scope for defence than in a classic dispute over content published on Facebook or Instagram. If the states succeed, the legal liability of the major platforms will become much more closely tied to the decisions of their product teams.
The judge can influence the code
The most far-reaching part of the process lies in the measures that states seek after a possible victory. The court could restrict itself to fines and narrow corrections to business practices, but state prosecutors are asking for much more. Their request reaches into the design of the product itself.
They want limits on how minors use the platforms, controls on features that prolong engagement, and different rules for the algorithms that determine what a young user sees.
Of particular interest is the request that Meta delete algorithms and AI models built using children’s data that, according to the states, were collected unlawfully.
Oakland may not provide a blanket rule for the entire AI industry
Such a measure would have consequences beyond Meta. An order to delete illegally collected data is relatively straightforward. With machine learning, however, a far more complex problem arises, because data can be removed from the database while the model trained on it remains a product of the process in which that data was used.
The court would then have to decide how far the remedy should go. If an illegally obtained dataset contributes to model training, is it sufficient to delete the original data, to retrain the model, or to withdraw the model altogether?
Oakland may not provide a blanket rule for the entire AI industry. If Gonzalez Rogers accepts even part of such a request, lawyers, regulators and companies will have a precedent that will be studied carefully far beyond social media.
The value of the data will then no longer depend only on whether the company is allowed to keep it, but also on what it has built with it.
Meta is already changing the experience for minors
The company is going to court with a very different product for young people than it had a few years ago. Teen Accounts already automatically limit content and contacts, turn off night-time notifications and introduce usage-time reminders.
In 2026, Meta expanded AI age estimation to automatically reclassify users judged by the system to be teenagers into protective mode, even when they had reported an older age. It introduced additional parental controls, stricter content settings, and new warnings related to self-harm and teenage conversations with Meta AI.
A much more important debate will concern how far the state can mandate changes
Those changes in themselves do not prove the state prosecutors' claims. Any large platform would in any case adjust its protection of minors under pressure from users, regulators and the public, but they show how technically feasible it is to modify what has been presented for years as the standard social network experience.
Night-time notifications can be turned off, usage time limited, minors automatically given a different feed, age estimated more accurately than simply entering a date of birth, and parents given more control.
This will make it difficult for the court to accept the argument that certain protective changes are technically unfeasible. A much more important debate will concern how far the state can mandate such changes, and what evidence it must have to justify intervening in product design.
Business risk is no longer measured only by the penalty
For Meta, the potential financial exposure is enormous. The company estimated in court filings that the calculation method advocated by the states could result in penalties of about $1.4 trillion.
The states have not publicly requested a specific final amount, so the figure should be understood as Meta’s estimate of the maximum possible exposure, not as the expected verdict.
Several thousand similar proceedings are under way in the California state system
A far more significant risk lies in the possibility that thousands of other plaintiffs will gain a legal pathway to follow. More than 3,000 federal cases brought by individuals, school districts and other plaintiffs against Meta, Snap, TikTok and Google have already been centralised before Gonzalez Rogers. Several thousand similar proceedings are under way in the California state system.
One verdict will not automatically resolve all those cases. It can, however, show which legal theories survive, what kinds of evidence convincingly link the design to the harm, and which features of the platform the court treats as business decisions of the company rather than publishing decisions protected by Section 230. That is far more valuable than a single large fine.
Congress may be slow, but the courts will not be
The US Congress has spent years trying to find a politically viable national online protection regime for children. Meanwhile, states have begun enacting their own laws, prosecutors have opened cases, and courts are now tasked with applying existing consumer protection, privacy and product liability rules to technology that emerged much later. Oakland can speed up that process.
The most likely outcome is not a court order that overnight remakes Facebook and Instagram across the United States. Whatever happens at first instance, Meta will appeal.
Congress may not write a detailed law dictating what Instagram should look like, while courts, on a case-by-case basis, may render certain aspects of its design too expensive to maintain
Section 230 and the First Amendment continue to place severe limitations on courts when a dispute involves the ranking and distribution of content. The April decision of the Ninth Circuit Court of Appeals showed how strong that protection remains.
Greater change comes gradually. Features that can easily be described as the platform’s own design, but hardly as editing someone else’s speech, will carry increasing legal risk.
For minors, the first elements to disappear or change will be those that prolong uninterrupted use, encourage them to return to the application, and make it difficult to exit.
Companies will introduce these changes before final verdicts, once they conclude that the cost of keeping the old feature is greater than its contribution to user engagement.
If that line is solidified, American regulation of social networks will take a form that the tech industry has not long anticipated. Congress may not write a detailed law dictating what Instagram should look like, while courts, on a case-by-case basis, may render certain aspects of its design too expensive to maintain.
The process beginning on 18 August should therefore be followed closely to see which corporate decisions the court considers to be its own responsibility.
That will determine how deeply the law can penetrate the code of products used by hundreds of millions of people every day.
