Meta agreed on Wednesday to pay states at least $12.19 billion over ten years and rebuild how teenagers use Instagram and Facebook, ending a trial in Oakland eight days after it opened.
The figure in every headline is $17.1 billion, the figure Meta is actually obligated to pay is $12.19 billion, and the gap between them is not the only thing in this deal that hangs on what three other companies decide to do.
Coverage of the settlement has read that contingency as a money term: Meta pays the remaining $5 billion or so only if TikTok, YouTube and Snapchat reach comparable deals with the states. Fortune went further into the structure than most outlets did and still described Meta’s safety obligations as fixed, with only the payout floating. The attorneys general’s own releases say something different. The Connecticut settlement summary and the District of Columbia’s announcement both describe a protection regime with two tiers. Teenagers get a combined two-hour daily cap across Instagram and Facebook, and that cap runs for five years. If Snapchat, TikTok and YouTube sign comparable terms, the cap drops to sixty minutes per platform and runs for ten. California Attorney General Rob Bonta’s office put it in one line: if other social media platforms agree to similar terms, the daily time limit will drop to one hour.
So the strongest child-safety provision in the largest child-safety settlement any American regulator has ever extracted is conditional, and the condition is litigation the states have not yet won.
What the States Actually Signed
Strip out the contingency and here is what Meta is committed to, with the durations the attorneys general themselves attached:
- A default two-hour daily limit across Facebook and Instagram for users under 18, removable only by a parent, with mandatory pauses at 15 minutes of continuous use and again at 60 and 90 minutes. Five years.
- A nighttime block from midnight to 6 a.m., again parent-removable, with notifications silenced from 10 p.m. to 7 a.m.
- No push notifications on weekdays between 8 a.m. and 3 p.m. during the school year.
- Age assurance measures to identify users under 18 and remove users under 13.
- Hidden like counts, a ban on cosmetic-procedure filters, an option to turn off the personalized feed, and a requirement to answer 90 percent of content reports within six hours.
- An independent auditor with access to Meta’s internal data, reporting to the attorneys general. Five years.
Read that list next to what Instagram already shipped in September 2024 and the shape of the deal gets clearer. Teen Accounts already muted notifications from 10 p.m. to 7 a.m. and already nudged teenagers with a notice after 60 minutes. The settlement converts a nudge into a cap and a mute into a block, which is real, and it narrows the hard-blocked window to six hours inside a nine-hour quiet period Meta had set for itself two years ago. The consequential change is the auditor, because for the first time somebody outside the company gets to check the homework.
“Meta strip-mined the souls of America’s children for maximum profit with abusive and addictive features that unleashed a youth mental health catastrophe.”
That is Connecticut Attorney General William Tong, describing conduct Meta has not admitted and that this settlement now closes the book on for 47 states, the District of Columbia and several territories.
Why Meta Would Rather Pay Than Cap Alone
The money was never the problem. Meta reported $200.97 billion in revenue and roughly $60.5 billion in net income for 2025. The guaranteed portion of this settlement works out to about $1.22 billion a year, which is two percent of a single year’s profit, paid annually for a decade. Set the whole guaranteed sum against 2025 earnings and it comes to roughly 74 days of net income, spread across ten years.
What Meta could not accept was capping teenage engagement while its competitors did not. An hour a day on Instagram is a very different product when TikTok is uncapped, and every minute a 15-year-old does not spend on Instagram is a minute available to somebody else. That is the asymmetry this settlement is built around, and it explains a clause that otherwise makes no sense: Meta wrote a deal in which both the money and the strictness of the child protections ratchet only when its rivals are bound by the same terms.
The states took that deal, and in taking it they accepted Meta’s framing of its own industry. Tech litigator Jess Nall told Fortune the contingency reinforces an argument Meta has made all along, that causation cannot be pinned on any single platform because teenagers use several of them. Meta now has a document signed by 52 attorneys general that treats that argument as an operating assumption.
New Mexico Tried the Case and Got More
The comparison that matters is not Meta against the wires. It is Meta against the one state that refused to settle and put the evidence in front of a jury.
New Mexico won the first jury verdict against Meta over child safety anywhere in the country, and earlier this month a court ordered Meta to pay $567 million into an abatement fund on top of the $375 million the jury had already awarded. Attorney General Raul Torrez told Fortune this week that the multistate deal does not go as far as what his state secured on its own, pointing to a direct ban on romantic and sexualized AI chatbot interactions with minors and stronger protections against adults contacting children in private messages. Neither appears in the agreement 51 other attorneys general just signed. New Mexico, Florida and Texas all sat out the multistate settlement, though Texas cut its own for more than $1 billion.
We covered the opening of the Oakland trial when the states were seeking damages nearly the size of the company, and the March verdict that established these cases can be won in front of a jury. The through line is not complicated. Litigation produced the unconditional remedies. Negotiation produced the conditional ones.
The Part the States Got Wrong
Our view: the attorneys general should not have let the strength of a child-safety remedy be priced as a competitive term.
Every other provision in this agreement rests on the premise that Meta’s design harms children. If a sixty-minute cap is the level of protection children actually need, it needed to be the cap regardless of what Snapchat does. If two hours is sufficient, then the sixty-minute tier is a bargaining chip dressed as a safeguard. It cannot be both. By writing it as a contingency, the states have built a world where a teenager’s screen-time protection in 2029 turns on the outcome of enforcement actions against companies that were never in the Oakland courtroom, and they have given Meta a standing interest in how those cases resolve. The cleaner design was available, and New Mexico demonstrated it: try the case, take the verdict, collect remedies with no escape hatch.
There is also the matter of age assurance, which every one of these protections depends on. As TechCrunch pointed out, the settlement sets no accuracy benchmark for determining who is under 18, which means the two-hour cap reaches exactly as many teenagers as Meta’s detection systems manage to find. Australia hit the same wall with its under-16 ban, where Meta deleted 550,000 accounts and nobody established how many it missed.
Judge Yvonne Gonzalez Rogers still has to approve the consent judgment, and Bonta says the changes arrive within months. The number worth watching is not $17.1 billion, and it is not $12.19 billion either. It is whether TikTok and YouTube ever sign, because until they do, the strongest protection in this settlement exists only on paper.