Instagram will cut teenagers off at midnight! Regulators everywhere now have a template

The money in Meta's settlement with American state attorneys general will be argued over for years. The product terms may matter more, and they are unusually specific.

Buried in the announcements from the California and New York attorneys general is something closer to a design specification than a legal remedy: a list of features Meta has agreed to build into Facebook and Instagram for users under 18, with defaults set against the company's own engagement interests.

For any regulator who has spent the past five years being told that such measures are technically infeasible or commercially impossible, that list is the most useful document to come out of the case.

What the specification requires

Under-18 accounts get a default daily limit of roughly two hours across Facebook and Instagram combined. California's summary notes the cap drops to one hour if other major platforms adopt comparable limits.

Access is blocked entirely between midnight and 6 am. Push notifications are barred between 10 pm and 7 am, with California describing a further block during school hours.

Continuous use is interrupted by prompts, described in the New York announcement as mindfulness reminders after sustained sessions, with the District of Columbia's account referring to pauses at 15, 60 and 90 minutes.

Like and reaction counts are hidden for under-18 users. Cosmetic procedure filters are restricted for minors. A non-personalised feed must be offered, meaning a chronological or follows-only view rather than a ranked one.

Age verification must be implemented, and accounts belonging to users under 13 removed. Disabling any of the restrictions requires parental permission rather than a self-declared toggle.

Why the defaults are the point

Almost every element on that list already exists somewhere as an optional setting. Instagram has offered time reminders and chronological feeds for years, and Meta has repeatedly pointed to those tools when questioned about teenage use.

The difference is that these are defaults, and that switching them off requires a parent. Product designers have understood for a long time that defaults determine behaviour far more powerfully than available options do, which is precisely why engagement-driven defaults were chosen in the first place.

The two-hour cap is the clearest case. A voluntary two-hour reminder is a nudge. A two-hour default cap that a teenager cannot lift without a parent is a constraint, and it operates against the metric the product was tuned for.

The infeasibility argument just got harder

The reason this matters beyond the United States is procedural rather than legal.

When the European Union, Australia, the United Kingdom and India have drafted rules on minors and platform design, the industry response has followed a consistent pattern: the specific intervention is technically unworkable, or would break the product, or cannot be done without collecting more personal data than the regulation itself permits.

Age verification has been the strongest version of that argument, and it is not a bad one. Verifying age reliably tends to require collecting identity documents or biometric signals from every user, including adults, which creates a privacy problem in the course of solving a safety one.

That argument survives this settlement. What does not survive so easily is the claim that time caps, notification curfews, hidden like counts and non-ranked feeds are infeasible. Meta has now agreed to ship them.

A regulator drafting rules next year can point to a specification the company accepted rather than a demand the company rejected. That is a materially different negotiating position.

The limits of the template

The strongest counter-argument is that a specification is only as good as its enforcement, and this one has three soft edges.

The first is age verification. Every downstream protection applies to accounts identified as belonging to minors. If a teenager registers as an adult, none of it engages. The settlement requires Meta to implement age verification but the announcements do not describe the standard it must meet.

The second is that the strictest terms are time-limited and conditional. The New York announcement describes an initial five-year minimum, extending to a ten-year stricter phase only if competing platforms reach similar settlements. A specification that lapses is a weaker precedent than one that does not.

The third is that compliance can be narrow. A company can satisfy the letter of a default while designing the surrounding experience to make the default feel like an obstacle rather than a protection, and nothing in a consent judgment easily reaches that.

There is also a substantive objection to the whole approach. Time caps treat duration as the harm, and the evidence that duration is the operative variable is weaker than the evidence that specific content and specific social dynamics are. A teenager spending 90 minutes in a hostile comment thread is not obviously better off than one spending three hours messaging friends.

What Meta has and has not conceded

It bears repeating that the settlement contains no admission of liability. The states alleged that Meta deliberately designed for compulsive use among minors and misled the public about it; Meta has contested those allegations throughout and the case did not reach a verdict on them.

Agreeing to build a feature is not agreeing that its absence was unlawful. Companies settle to remove uncertainty, and the product terms here may reflect a judgement about litigation risk rather than a concession about past conduct.

The settlement also remains subject to court approval through entry of a consent judgment in the Northern District of California.

What to watch

The first real test is the age verification implementation, because everything else depends on it. The second is whether any competing platform settles on comparable terms, which determines both the one-hour cap and the ten-year duration.

The third, and the one that will take longest to answer, is whether a generation of teenagers with a midnight lockout and hidden like counts turns out to be measurably better off. The settlement assumes the answer is yes. It does not prove it.