What was agreed
Meta and a bipartisan coalition of state attorneys general filed a proposed settlement Wednesday morning in the federal child safety case being tried in Oakland before Judge Yvonne Gonzalez Rogers. The court filing puts the payment at $16.68 billion. Rob Bonta’s office says $17 billion, and several attorneys general have used $17.1 billion, a number that folds in $459 million resolving separate state privacy claims from the Cambridge Analytica affair. Meta’s own newsroom post says roughly $18 billion, paid in annual installments over ten years. The figures differ for two reasons: what each party counts as part of this case, and the fact that a large chunk of the money is conditional. That condition is possibly the most interesting part of the deal.
Whichever number you take, it is spread across ten years and refers to a company that booked $60.8 billion of revenue in a single quarter this year and has guided to spending between $130 and $145 billion on infrastructure in 2026 alone. The largest child safety settlement in the industry’s history comes to about six weeks of Meta’s capital expenditure.
Why Meta settled when it did
Meta settled midway through Adam Mosseri’s testimony and before Mark Zuckerberg was scheduled to take the stand.
The court time that preceded the settlement produced a steady run of damaging testimony. Arturo Béjar, who spent eight years working on safety at the company, testified that safety was never a meaningful priority and that Zuckerberg’s 2021 public denial of putting profit ahead of wellbeing was, well, inaccurate. He called autoplay, like counters and infinite scroll inherently unsafe for teenagers, and described Meta’s wellbeing features as designed to fail because they ship as optional settings. His great analogy was that you have to turn on the airbag every time you get into the car. The states then put Mosseri on the stand and asked him why the company publicised Take a Break when internal uptake among teens was negligible.
Settling cost Meta a large sum of money. Continuing would have cost it another six weeks of testimony like these and would have seen Zuckerberg on the stand too with potentially more damaging results.
Dragging everyone else in
Roughly 70 percent of the payment, about $12.7 billion, is guaranteed to the participating states over the decade. The remaining 30 percent, about $5.3 billion, is released only if competitors adopt a matching framework covering daily time limits, night mode and age assurance, and pay matching sums. Meta’s own account names TikTok and YouTube; Reuters reported the contingency covers Snap as well. Meta’s commitments run for five years, extending to at least ten if its peers sign up.
Chief legal officer C.J. Mahoney’s framing is that teens move fluidly across dozens of apps, so the solution has to be industry-wide. That is true as a description of teenage behaviour. But this is not the real reason. It is also a description of Meta’s competitive problem. A company that unilaterally caps teen usage at two hours a day hands those hours to whoever has not capped them. Making the limit universal removes the penalty for complying, converts a Meta-specific liability into an industry-wide cost, and positions Meta as the author of the standard rather than its target. Meta is making sure that this is understood by taking out an ad in the Washington Post, the LA Times and the NY Times on Thursday morning as reported by Mike Isaac. Meta’s stock rose on the announcement. Snap’s fell more than 8 percent. The market understood the trade immediately.
There is a second benefit. Congress has been sitting on the Kids Online Safety Act for years. A negotiated framework, already agreed with 51 attorneys general, is now the obvious template for whatever Washington eventually passes, and Meta is holding the pen.
What Meta has agreed to change
Under-18 accounts get a default daily time limit of two hours that only a parent can lift. Overnight use is blocked between midnight and 6am. Notifications are muted at night and during school hours. Like and reaction counts are hidden on minors’ posts. Cosmetic surgery and extreme makeup filters are banned for minors. Young users get the option of a non-personalised feed. Parental supervision tools are expanded. Meta will implement age assurance to detect under-18s and remove under-13s. An independent auditor gets access for five years, and Meta is enjoined from making further false or misleading claims about its safety features.
Almost all of that already exists at Meta in some form, sitting behind a toggle that teens do not flip, which is precisely what Béjar described on the stand using the airbag analogy. What the settlement changes is the default position of the switch, plus an auditor and an injunction to make the switch stay put. Moving these settings from optional to default can have a positive impact, but it alo strikes me as the bare minimum of what could count as real change. Nothing in the agreement turns off recommendation algorithms by default, and the structure leans on giving parents tools rather than removing the features underneath.
Every default depends on age assurance
Correctly identifying who is a minor has been the unsolved problem for two decades. The COPPA claim in this very case exists because under-13s registered with false birthdates. Kids are industrious in trying to fool the age estimation software, including using makeup and other tactics to appear older. Australia, the UK and France have all moved toward under-16 restrictions this year and hit the same wall.
The Center for Democracy and Technology, whose Free Expression Project director Kate Ruane gave a statement to Fortune, raised the cost that comes with the fix: establishing that a user is a child requires establishing the age of every user. Meta has just committed to that at the scale of Facebook and Instagram, under a consent judgment, with an auditor checking. Whatever else this settlement produces, it normalises identity verification as the price of using a social platform.
The part I am not celebrating
Device makers have been shipping Screen Time, Digital Wellbeing and Family Link for the better part of a decade. Parents have had timers and blocks and reports that whole time, and compulsive use did not fall. The documented damage to self-esteem, sleep, body image and friendship kept accumulating. Tools handed to families have never been a match for products engineered to defeat them, and the adults in this story are not doing measurably better than the teenagers.
Meta is paying, and paying to avoid a verdict in a case where the states’ demand ran into the hundreds of billions. It accrues roughly $10 billion of legal expense in the third quarter and moves on. Two numbers matter more than the headline: five years, which is how long the commitments are guaranteed to last, and the third of the money that arrives only if competitors move first.
States have now established a floor for what accountability looks like. Whether that floor becomes law, and whether it covers the design choices rather than the parental dashboards, is a decision Congress has deferred for years and Meta has just gained considerable influence over.