Instagram expands AI teen detection globally, even for accounts with adult birthdays

Meta said on November 9 that it is expanding AI age estimation worldwide on Instagram, moving suspected teenagers into Teen Account protections even when they registered with an adult birthday.

Teen Accounts already exist across Instagram’s markets: Meta introduced them in 2024 and announced worldwide availability in June 2025. The change broadens detection of teenagers who supplied an adult age. Proactive AI classification began testing in the United States in 2025, followed that year by the United Kingdom, Canada and Australia. That earlier announcement also offered a way to change settings when the system made a mistake. Those protections also require parental permission for under-16s to go Live or turn off the shield hiding unwanted images in direct messages.

Meta’s May technical explanation describes looking across profiles for contextual clues, including birthday celebrations and references to school grades. Posts, comments, biographies and captions can contribute signals. The assessment therefore draws on information left through account activity as well as the birthday entered during registration, according to the company’s description.

Photos and videos provide additional visual signals. In a July correction, Meta said this analysis recognizes general patterns associated with younger people and does not use biometric data to identify a person or create facial fingerprints. Its May post limited visual analysis to selected countries; the new announcement supplies neither country-level coverage for that technique nor new error-rate figures.

Teen Accounts restrict messages to followed or connected users and tags to followed users. Filters screen offensive comments and message requests. Private defaults cover all under-16s and new under-18 users. New followers need approval; non-followers cannot view private content.

Age also determines control over settings. A November 6 Family Center explanation by ConnectSafely CEO Larry Magid says 13-to-15-year-olds need parental permission to weaken protections. Older teenagers can change some settings, while content protections still require a parent’s permission. Suspected under-13 accounts face a separate action: deactivation, followed by a requirement to prove age to prevent deletion. The guide describes age assurance as a range of methods, from asking for a birthday to technical estimation and, in some cases, checking identification or facial age estimates.

Australia adds Apple’s age-range signal

Meta said it has begun using Apple’s age-range signal in Australia.

Apple’s developer documentation describes an API that returns a range with lower and upper bounds, keeping the exact birthday private. Parents or guardians of children in Family Sharing can choose ongoing sharing, refuse it or decide app by app. Apps specify age thresholds, and the system returns the corresponding range. Apple tells developers to treat that response as a range rather than infer an exact age. The data is based on information declared by a user, parent or guardian. After an age boundary is crossed, a cached range can persist until the declaration’s anniversary; people can update it in settings.

Apple’s separate developer Q&A distinguishes legally defined categories in some regulated regions from ranges supplied elsewhere when people choose to share. Developers remain responsible for their apps’ age restrictions. Since iOS and iPadOS 26.5, when an age range has been confirmed with identification or a payment method, the API reports a single confirmed value without disclosing which of those methods was used.

Meta’s Facebook rollout continues, while a similar global launch on Threads is planned for later this year.