Meta is introducing a new artificial intelligence technology that analyzes users’ bone structure to identify children under 13 on Facebook and Instagram. The innovative approach is part of the company’s efforts to comply with strict online child protection regulations.
See also: Meta tests premium subscriptions on Instagram, Facebook and WhatsApp
According to a statement from Meta, the new AI will scan photos and videos posted to its platforms for “general themes and visual cues,” including height and bone structure. The company emphasizes that “this is not facial recognition” and does not identify the specific person in the image.
The technology represents a major development in the field of protecting minors on social media. The system will also analyze posts, comments, bios and captions to look for “contextual evidence” that may indicate that someone is a minor. The technology is initially available in “select” countries, including the US, before a wider rollout.
Meta 's new technology relies on machine learning algorithms that analyze body proportions and growth characteristics to estimate users' ages. Unlike facial recognition, which has faced bans in many European countries, bone structure analysis focuses on general physical characteristics that do not reveal a person's identity.
See also: Meta – Global problems on Facebook, Instagram and WhatsApp
Initial test results show that the system achieves 85-95% for the 4-12 age group by analyzing height and bone structure ratios. However, researchers warn of potential accuracy issues across ethnicities, with an error rate that can reach 15% for non-Caucasian children.
Accounts identified as underage will be automatically disabled, and the account holder will be required to verify their age to avoid deletion. This process is part of Meta to comply with COPPA (Children's Online Privacy Protection Act) and other child protection regulations.
The new technology has drawn mixed reactions from experts and privacy groups. Jennifer Granick of the Electronic Frontier Foundation (EFF) called it “privacy theater,” arguing that bone structure analysis still indirectly infers identity.
On the other hand, Meta CTO Andrew Bosworth defended the technology as “ essential for faceless security ,” claiming 90% accuracy in tests. James Steyer of Common Sense Media praised the enforcement capabilities but warned against excessive account removal.
See also: EU: Research in Meta on negative effects of Instagram, Facebook on children
The child safety technology market, part of the $50+ billion AI ethics and mitigation sector , is expected to reach $75 billion by 2028. Competitors like TikTok are already using similar technologies, with Snapchat implementing posture analysis AI by 2024
