Meta is expanding Instagram AI age estimates worldwide, including for accounts that list an adult birthday

Documented: On November 9, 2026, Meta said it is expanding AI age-estimation on Instagram globally so that accounts it suspects belong to teenagers are placed in Teen Account settings even if the registered birthday is adult. The company said the change should put millions more teens into those protections, which automatically limit who can contact them, what content they see, and time on the app. U.S. parents are to receive an Instagram notice this month explaining the process. Meta said the same approach continues rolling out on Facebook and is planned for Threads later this year. In Australia, Meta said it has begun using Apple’s optional age-range signal, passed by the App Store when a parent has set an age range on a teen’s device.

Documented context, not a new allegation: Meta ties the expansion to an August 2026 agreement with U.S. state and territorial attorneys general. Reuters, state AG offices, and court-related reporting described settlements totaling up to about $18 billion — roughly $16.7–17.1 billion across a multistate deal covering 47 states plus D.C. and several territories, and a separate Texas deal of about $1 billion — resolving claims that Instagram and Facebook were designed in ways that harmed children. Those terms include time limits and other teen safeguards and commitments to improve identification of under-13 accounts and of teens who enter an adult birthday. Meta’s own November 9 post says the global Instagram rollout builds on that agreement and on earlier tests in the United States, the United Kingdom, Canada, Australia, the European Union, and Brazil.

Claimed by Meta, not independently audited in the public announcements: the estimator looks beyond the birthday field at profile text, posts, comments, captions, and bios (for example school grades or birthday mentions) and at general visual patterns in photos and videos to estimate an age range. Meta and internet-safety advocate Larry Magid, writing on a Meta family resource page the company pointed parents to, say that visual step is not facial recognition and is not meant to identify a person. Accounts suspected to be under 13 are deactivated pending proof of age. Teens 13–15 need parental permission to loosen some protections; 16- and 17-year-olds have more control but still face limits. Meta says people placed in Teen Accounts by mistake can challenge the result, including with ID or a third-party video age check.

The privacy tension is structural, not speculative rumor. Age assurance that ignores a self-reported adult birthday necessarily scans behavior and imagery of people who may be adults. Meta argues a single device-level signal shared with many apps is less invasive than every platform collecting IDs; it also says app-store sharing of that signal remains optional, so coverage will be uneven. Separately documented the same week: Meta, Google, and TikTok sued to block California’s AB 1709, which would bar personalized feeds and autoplay for users under 16 and require account deletion if those features stay on, with penalties up to $50,000 per affected user. The companies argue the law violates the First Amendment and is preempted by Section 230. That suit is a legal challenge to a state design rule, not evidence that Meta’s own estimator is accurate or inaccurate.

Speculated, and not established by Meta’s posts or the settlement announcements: error rates for adults misclassified as teens, how long images or signals are retained beyond what Meta has already described for ID and selfie checks, and whether the system will be used for purposes other than Teen Account placement. Those gaps are the live dispute — parental control versus inference over people who say they are adults — and they are not resolved by the company’s claim that the rollout is working.