Meta has deployed AI systems that analyze photos on Instagram and Facebook to identify minors based on physical characteristics like bone structure and body size. The company explicitly avoids facial recognition, instead relying on body composition analysis to flag accounts belonging to children.
The system serves Meta's age-verification efforts on platforms where the company faces regulatory pressure to protect minors. Rather than asking users to provide documents, Meta uses computer vision to estimate age from skeletal and physical markers visible in uploaded images.
This approach sidesteps some privacy concerns around facial recognition but introduces different questions. Analyzing bone structure and body measurements raises accuracy concerns, particularly across different populations with varying physical development patterns. The AI could misclassify adults with certain body types or misidentify adolescents.
Meta's emphasis on avoiding facial recognition reflects regulatory backlash against the technology. The EU's Digital Services Act and similar rules globally push platforms toward age verification without relying on facial biometrics. Meta's solution uses what it frames as less invasive markers.
The company hasn't disclosed how accurate this system performs or whether it's been independently audited. Meta also hasn't fully explained how it handles false positives, which could result in young-looking adults being flagged as minors or vice versa.
The broader context matters here. Meta faces constant scrutiny over child safety on its platforms. This detection system could support age-gating of certain content or restricting data collection for younger users. But it also means Meta is analyzing body data at scale from millions of uploaded photos, creating a new category of biometric surveillance.
The practical impact remains unclear. Detection alone doesn't solve protection. Meta would need enforcement mechanisms to prevent minors from lying about their age or to stop adults from accessing child-focused accounts.
THE BOTTOM LINE: Meta's body-analysis approach to age detection avoids facial recognition criticism but creates new privacy tradeoffs and unproven accuracy claims