AI age estimation expands beyond the initial rollout
Meta is expanding its AI age-estimation technology globally to identify people it suspects are teenagers even when an account uses an adult birthday. Those users will be placed into Instagram Teen Account protections, which limit who can contact them, the content they see and time spent in the app. Meta says millions more teens will receive the settings.
Age assurance is difficult because a self-declared birthday can be inaccurate, while identity-document checks collect sensitive data from everyone. Meta’s approach uses behavioural and other signals as a backstop. The company says it will continue the rollout on Facebook and begin launching similar technology globally on Threads later this year.
Teen protections depend on classification quality
An age-estimation system has two important error types. Missing a teenager can leave an account without appropriate safeguards, while classifying an adult as a teen can restrict a legitimate user. Meta needs an accessible correction and appeal process for both cases, with response times appropriate to the impact on communication and content access.
Evaluation should cover countries, cultures and patterns of use, not only aggregate accuracy. Signals that correlate with age in one group may perform differently elsewhere. Independent testing and transparent error ranges would help families and regulators understand how reliably the system works. High coverage is valuable only when mistakes can be identified and corrected.
Australia adds Apple’s age-range signal
Meta has begun using Apple’s age-range signal in Australia as another way to understand age. When a parent configures a device for a teenager, the app store can confirm an age range to a developer when an app is downloaded. Meta argues that one device-level signal shared with apps is more privacy-preserving than every service collecting its own identity evidence.
The signal is optional for app stores, developers and families, so it cannot be the only method. Device sharing and incorrect family settings can also produce errors. Combining it with Meta’s estimation may improve coverage, but the product should explain which source triggered a restriction and how a parent or user can correct inaccurate information.
Parents will receive more explanation
Parents in the United States will begin seeing an Instagram video explaining how Meta places teens into age-appropriate protections and how families can help keep age information correct. The material links to a broader explanation of Meta’s layered approach, including efforts to keep children under 13 off its services.
Clear communication matters because invisible classification can feel arbitrary. Parents and teenagers should know what protections change, what data contributes to the estimate and what happens during an appeal. Explanations need to be understandable without revealing enough detail for bad actors to evade the system deliberately.
The rollout sits inside a broader regulatory debate
Meta says the expansion builds on an agreement with 52 bipartisan attorneys general and calls for industry peers to adopt similar practices. It also supports app-store age verification and parental-consent legislation, arguing that consistent device signals could reduce invasive checks by individual platforms. Governments worldwide are introducing youth-safety rules but often leave implementation details unresolved.
A shared signal can improve consistency, yet it concentrates responsibility in device and store ecosystems. Standards need privacy limits, security protections and a way for users to challenge incorrect data across services. Legislation should define outcomes and accountability without forcing every person to provide a government identity document for ordinary online access.
Safety gains require ongoing evidence
Automatically applying Teen Account settings can reduce exposure to unwanted contact and unsuitable content, particularly when a young person misstates their birthday. The system should be monitored for evasion, bias and unintended restrictions as behaviour changes. Meta should publish evidence on corrections, appeals and the number of accounts moved into and out of protection.
The global expansion is a significant use of AI for platform governance rather than content generation. Its legitimacy will depend on proportional data use and meaningful user control. Age estimation can support safer defaults, but it should remain one layer alongside parental tools, product design, reporting and firm action against adults who target young users.
Children under 13 require a separate response
Meta says it is strengthening technology to identify accounts that may belong to children under 13, alongside protections for users aged 13 to 17. These groups have different legal and product requirements. A system should not simply place a younger child into a teenager experience and consider the problem solved. Detection must lead to the correct account and parental process.
Signals used for age assurance can also become targets for manipulation. Meta needs red-team testing that explores coordinated evasion, purchased accounts and adults pretending to be teenagers for harmful contact. Defensive models should adapt without making the appeal process impossible for legitimate users. Safety requires both accurate classification and strong enforcement against abusive behaviour online.