Instagram is beginning to test the use of artificial intelligence to determine if kids are lying about their ages on the app, parent company Meta Platforms said on Monday.

Meta has been using AI to determine people's ages for some time, the company said, but photo and video-sharing app will now “proactively” look for teen accounts it suspects belong to teenagers even if they entered an inaccurate birthdate when they signed up.

If it is determined that a user is misrepresenting their age, the account will automatically become a teen account, which has more restrictions than an adult account. Teen accounts are private by default. Private messages are restricted so teens can only receive them from people they follow or are already connected to. “Sensitive content,” such as videos of people fighting or those promoting cosmetic procedures, will be limited, Meta said. Teens will also get notifications if they are on Instagram for more than 60 minutes and a “sleep mode” will be enabled that turns off notifications and sends auto-replies to direct messages from 10 p.m. until 7 a.m.

Meta says it trains its AI to look for signals, such as the type of content the account interacts, profile information and when the account was created, to determine the owner's age.

The heightened measures arrive as social media companies face increased scrutiny over how their platform affects the mental health and well-being of younger users. A growing number of states are also trying to pass age verification laws, although they have faced court challenges.

Meta and other social media companies support putting the onus on app stores to verify ages amid criticism that they don’t do enough to make their products safe for children — or verify that no kids under 13 use them.

Instagram will also send notifications to parents “with information about how they can have conversations with their teens on the importance of providing the correct age online,” the company said.