Meta announced on Tuesday that it will introduce an AI-based visual analysis system to scan photos and videos on its platform to identify users who may be under 13 years old, and decide whether to remove their accounts on Facebook and Instagram. The company stated that the system estimates age by analyzing general visual features such as height and bone structure, and combines text content and interaction behavior for comprehensive judgment, to enhance the ability to identify and manage minor accounts. It also emphasized that this technology does not involve facial recognition or specific identity identification.

Meta noted that the system has been launched in some countries and is gradually expanding its coverage. As part of its strategy to protect minors, the company will also use semantic clues in user profiles, posts, comments, and bios (such as birthdays and academic-related content) to assist in determining age, and plans to extend the relevant capabilities to more scenarios such as Instagram Live and Facebook groups. Once the system determines a user may be below the age threshold, their account will be suspended, and the user must go through the official age verification process to appeal and restore their account.

Meanwhile, Meta is further advancing its "Teen Account" mechanism, expanding its deployment to 27 countries in the EU and Brazil. It provides stricter default privacy and content controls for teenagers, including receiving only private messages from followers, hiding potentially harmful comments, and default private account settings. This mechanism will also be introduced to the U.S. version of Facebook for the first time and is planned to expand to the UK and EU regions in June.

This upgrade comes amid increasing regulatory pressure. Weeks ago, a jury in New Mexico ruled that Meta must pay a $375 million civil fine for misleading about platform safety and endangering children's safety, and required it to make structural changes. Overall, Meta is strengthening its minor protection and compliance capabilities by introducing multimodal AI identification and a tiered account system to address increasingly stringent regulatory environments and legal risks.