Meta is expanding how it uses AI to detect teens on Instagram — and, in some cases, will begin overriding account settings.

Instagram announced it was using AI for age detection in 2024; the system looked for signals that users were under 18 years old, like if messages from friends say “happy 16th birthday,” for example (Meta also says it uses engagement data — people in the same age group often interact with content they see in similar ways). Teen accounts on Instagram are subject to more restrictive settings: by default, teen accounts are private, strangers aren’t able to send them messages, and Instagram limits what kind of content teens see. Last year, Instagram changed the settings for all teens on the platform so that safety features were automatically enabled.

The company now says it will use AI to proactively look for teen accounts that have an adult birthday, and change settings for users it suspects are kids. In a blog post, Instagram says it will begin testing the feature today in the US. If it detects that a user is actually a child but the account says otherwise, Instagram will automatically place it under the more restrictive teen settings. Instagram acknowledges the possibility that the system will make errors — the company says users will have the ability to change their settings back.

Meta has gradually offered more settings that it says will protect teens and kids on the platform, often in response to concerns from parents and lawmakers. Last year, the European Union launched an investigation into whether Meta was doing enough to protect the health of young users. Disturbing reporting into predators on Instagram targeting kids prompted a lawsuit by a US state attorney general. There’s also disagreement among tech companies — namely Google versus Meta, Snap, and X — over who should be responsible for keeping kids safe online. In March, Google accused Meta of trying to “offload” its responsibility onto app stores following the passage of a bill in Utah.