OAKLAND, Calif. — The question at the center of the federal courthouse on Clay Street this week is not what Facebook and Instagram show teenagers. It is how the two products were built.
Opening arguments began Tuesday in a bipartisan case brought by 29 states against Meta Platforms, in which state lawyers argue the company engineered its flagship apps to maximize the time minors spend inside them, and then misrepresented to parents and the public what it knew about the consequences. California, Colorado, Kentucky and New Jersey lead the consumer-protection claims. All 29 states join a federal claim that Meta collected personal data from users under 13 without parental consent, in violation of the Children’s Online Privacy Protection Act.
“Meta’s business model can be summed up in four simple words: ‘hook’ the users, ‘hold’ them for as long as they can, ‘harvest’ their data, and then ‘hide’ the truth from the public when making public statements,” California Deputy Attorney General Megan O’Neill said in her opening statement before Chief U.S. District Judge Yvonne Gonzalez Rogers. “It was especially bad for kids.”
The states’ case leans on Meta’s own paper trail rather than on any individual user’s experience. O’Neill cited a 2016 internal email describing the “overall company goal” for Instagram as “teen time spent” on the platform, and an internal study titled “Long Term Retention: The Young Ones Are The Best Ones,” which found that the earlier someone joins, the longer they stay and the more revenue they generate. Executives, O’Neill argued, publicly downplayed the very dynamics their own researchers were measuring. “Meta said it prioritized safety over profits, but it hid the reality that when it came time to make a decision, time and again profits won,” she said.
The remedies sought are unusually concrete: the states want Gonzalez Rogers to order Meta to strip a list of features from the experience of users under 18 — infinite scroll, autoplay, visible “like” counts, beauty filters and recommendation algorithms tuned to lengthen sessions — and to impose limits on daily use, including blackout windows during school hours and overnight.
Meta rejects the framing and the arithmetic. The company has told the court its theoretical maximum exposure under the states’ penalty calculation runs to roughly $1.4 trillion, approximately its entire equity value on the Nasdaq, and calls that figure proof the case is untethered from the claims pleaded. “The State AGs may call this a landmark case, but their limited claims are unsubstantiated and their financial demands are vastly disproportionate,” a Meta spokesperson said the day before opening arguments. “The AGs offer no proof anyone in their states was misled, claim benign features like having an additional Instagram account somehow harmed their residents, and attempt to penalize Meta for industry-wide challenges like age verification.” The company points to teen protections it has built, including tightened default privacy settings and a one-hour Instagram timer.
California Attorney General Rob Bonta, whose office filed the complaint in October 2023 after a multistate investigation, said the states are not asking for a specific dollar figure and accused Meta of publicizing the $1.4 trillion number to make the suit look unserious. Pressed on what the right number would be, he gestured at the company’s scale: “They generated revenue of $200 billion last year, so, you know, maybe that amount would be appropriate. Maybe more. Maybe less.” His stated aim is structural. “We want Meta, in short, to stop hurting our kids, stop knowingly hurting our kids, stop using features that you know create mental health harms to kids,” he said.
Why it matters: the shift from content to architecture
For two decades, Section 230 of the Communications Decency Act and the First Amendment have functioned as a near-complete shield for platforms, on the theory that what a service hosts is somebody else’s speech. This trial routes around that shield. The states are not suing over posts. They are suing over product design — the ranking function, the session-extension mechanics, the notification cadence — and arguing these are engineering choices subject to ordinary consumer-protection law, no different from a defectively designed appliance.
If that theory survives, its reach extends well past social media. Recommendation systems are the same class of machinery now embedded in AI assistants, chatbot companions, algorithmic shopping feeds and personalized education products. A finding that optimizing an objective function for engagement among minors is itself an actionable design defect would establish that model behavior — not just model output — can carry liability. That is a far broader compliance surface than content moderation: it implicates what a system is trained to maximize, how it is A/B tested, and what internal research a company generates and then declines to act on. Meta’s discovery record is doing most of the work in Oakland; every AI company running engagement experiments is generating a comparable one.
The precedent risk is why TikTok, YouTube and Snap are watching. An injunction ordering age-gated feature removal would be nationwide in practical effect — no company maintains separate architectures for 29 states — and would arrive as a court-imposed baseline Congress has failed to legislate.
What to watch
The trial is expected to run six to eight weeks. The jury will return an advisory verdict only; Gonzalez Rogers retains final authority over liability, penalties and any injunction — meaning the remedy, not the verdict, is the headline. Mark Zuckerberg and Instagram head Adam Mosseri are both expected to take the stand. So is Arturo Bejar, a former senior Meta engineer turned whistleblower, whom the judge cleared to testify last week after dismissing Meta’s motion to exclude him as a “‘Hail Mary’ attempt to eliminate a strong witness for the plaintiffs.” Meta has already absorbed two adverse outcomes this year — a $6 million Los Angeles jury award in March and a $567 million New Mexico judgment with mandated safety changes — and is expected to appeal any loss here, potentially to the Supreme Court. The durable question is whether the design theory survives that appellate journey intact.
“Meta's business model can be summed up in four simple words: hook the users, hold them for as long as they can, harvest their data, and then hide the truth from the public when making public statements.”— Megan O'Neill, Deputy Attorney General, State of California