Meta is expanding its age verification tool by using AI to detect teen users and then automatically putting them in more restricted teen account settings, even if users enter their adult birthday.
The company said in a blog post on Monday that it will proactively identify accounts it suspects belonging to teenagers and transfer them to these more stringent environments. It acknowledges that the system may make mistakes and that users can adjust their settings if needed.
“We have been using AI to determine age, but leveraging it in this way is a big change,” Mehta said.
The move is part of a broader effort by META and other social media platforms to better protect younger users as parents and lawmakers continue to pressure on them, who believe tech companies are not doing enough to maintain the safety and mental health of teenagers.
The Instagram teen accounts launched last year include built-in protections that limit who can contact teenagers, what they can see, and how long they can spend on the platform. Last year, the company automatically provided these security features to all teen users, requiring parental or guardian approval from anyone under the age of 16 to make changes.
The company said about 97% of users under the age of 16 have secured these protections in place. It also recently expanded its teen account experience to Facebook and Messenger.
Meta said it will start sending notifications to Instagram parents and provide information about why teen discussions provide accurate age information online.
“Understanding the age of people online is a challenge for the entire industry,” the company said. “We will continue to work to help ensure that teenagers are in age-appropriate online experiences, such as teen accounts, but the most effective way to understand age is by getting parents’ approval and verifying age on the App Store.”
While some organizations praise these types of age verification methods, others, author Devorah Heitner, who grew up in public, called privacy the meaning of “fearing.”
“In order for AI to effectively determine user age, it must know more than it should, especially for new users with a limited digital footprint,” she said. “We need to see social applications protect all users more from intrusive algorithms and harassment rather than focusing on age.”
