OAKLAND, 27 AUG 2026 — Meta has settled the multistate youth safety case for up to about US$16.7bn, eight days into a trial scheduled to run six weeks. Teen accounts will default to a two-hour daily limit across Facebook and Instagram, the platforms will be blocked to under-18s between midnight and 6am, and notifications will be muted during school hours.
The structure of the payment is more interesting than its size. Only 70 per cent is unconditional. The remaining US$5.3bn depends on what TikTok and YouTube agree to do.
What was agreed
We reported yesterday on the trial itself and its advisory-jury structure. It opened on 18 August before Judge Yvonne Gonzalez Rogers and has now ended without a verdict.
Meta pays roughly US$12.7bn to the states over ten years. It accepts the product changes above, together with strengthened age assurance intended to keep younger children off the services altogether.
Twenty-nine states were plaintiffs at trial. The settlement is reported to resolve the claims of 51 attorneys general, which is the difference between the coalition that was in court and the far larger group whose claims are being retired at once. Judge Gonzalez Rogers must still approve it.
Thirty per cent of it depends on TikTok and YouTube
The contingency runs like this. Meta pays the final 30 per cent only if TikTok and YouTube also accept daily time limits, night mode and age assurance measures, with each of them paying a comparable sum.
This inverts the usual direction of a settlement. A defendant's liability is tied to its own conduct. Here a substantial part of Meta's payment depends on the conduct of two companies that are not party to the case.
What Meta has bought is insurance against unilateral disadvantage. Capping teen usage at two hours is a product decision with a cost, and that cost is far higher if competitors face no equivalent limit and absorb the displaced attention. The contingent tranche compensates Meta, in effect, for the possibility that it has been regulated alone.
It also does something less comfortable. It gives the state attorneys general a direct financial interest in extracting the same terms from TikTok and YouTube, and gives Meta a US$5.3bn interest in their doing so. The enforcement of youth safety rules across the rest of the industry now has Meta's money behind it. That may be efficient, and it is not obviously how public enforcement is supposed to work.
Meta's condition asks more of its rivals than it accepted
The asymmetry in the terms is easy to miss in the coverage.
Meta has agreed to a two-hour daily limit, measured cumulatively across Facebook and Instagram together. The trigger releasing the contingent payment reportedly requires TikTok and YouTube to adopt one-hour daily limits.
So the condition Meta has attached is stricter than the condition Meta accepted. Whether that reflects a judgement about the relative intensity of short-form video, or simply the negotiating position of a defendant paying to end a trial, is not something the public terms explain. A straightforward reading of the deal as levelling the playing field is not accurate. The one it describes tilts the other way.
A default limit is not a limit
The reporting describes teen accounts defaulting to the two-hour setting. Everything depends on that word, and the published terms do not say whether the default can be changed, by whom, or with what friction.
The distinction decides how much this changes. A cap a teenager can lift in settings is a nudge. Defaults are the most powerful lever in product design and most users never alter them, so it would be a substantial nudge, but a nudge nonetheless. Requiring a verified parent to lift it makes the measure closer to a control. A cap that cannot be lifted at all is a hard limit, and by some distance the most aggressive term in the agreement.
The same question applies to the overnight block and the school-hours notification mute. A block that a determined user can switch off addresses inattention rather than compulsion, and compulsion is what the states alleged.
This is not a reason to dismiss the remedies, and the published terms do not settle the question. Anyone reporting this as a mandatory two-hour cap on teenage social media has read a word that is not there yet. The implementation detail, when it is filed, is the part worth reading.
Why the totals differ depending on where you read
Reported figures range from US$16.68bn to US$18bn, and the state count appears variously as 29, 47 and 51. None of these is necessarily wrong.
The spread comes from what is being counted. A total that includes the contingent tranche is larger than one that counts only what Meta must pay regardless. Some figures appear to aggregate related settlements rather than this one alone. And the state numbers differ because the coalition at trial, the states whose claims the agreement resolves and the states receiving payments are three different sets.
The number to hold is the one with the fewest conditions attached. About US$12.7bn is what Meta pays if nothing else happens. Everything above that is a projection about two other companies.
What the settlement stops from being decided
A trial produces findings; a settlement produces terms. The difference matters for anyone who was watching this case for a precedent.
The questions the trial was going to test are now unresolved. Whether engagement-maximising design can be treated as a defective product, whether the evidence supports a causal link to adolescent harm, and how far a platform's knowledge of that harm can be established from its own documents all remain open. They were the reason this case mattered beyond Meta, because a ruling would have set the reference point for the theory itself.
What remains is the discovery record, which does not disappear, and a template of remedies that other regulators can point at. That is considerably less than a judgment, and Meta has paid for the difference.
The legal theory that got this to court remains untested there. The idea that design choices are actionable where content is protected, which the Ninth Circuit accepted when it declined to treat Section 230 as blanket immunity, will have to wait for the next case.
What it means from here
For the region, the operative part is the remedies rather than the money. A two-hour default cap, a night-time block and school-hour notification limits are now a documented, agreed, implementable set of controls that a company has accepted rather than litigated.
That makes them a reference. Regulators drafting rules elsewhere no longer have to argue about feasibility, because the largest operator has conceded it. Malaysia's age verification requirement for under-16s was drafted without that reference point, and whatever follows it will not be.
The thing to watch is not Meta. It is whether TikTok and YouTube accept comparable terms in the coming months, because US$5.3bn now rides on it, and because if they do not, the first serious limits on teenage social media use will apply to one company while its competitors carry on.
View author profile → · Editorial policy
