Meta AI Compiled Child Profiles From Family Posts, Now Fixed

Meta patches AI feature after viral privacy incident

Meta has corrected a significant privacy issue with its AI chatbot after the tool proactively prompted a user to investigate her own child's identity, then assembled detailed profiles of her two daughters by aggregating years of content posted by family members across Facebook and Instagram.

The incident came to light when Kalie Robins, a Utah-based Instagram user, cross-posted a car video featuring one of her daughters to Facebook. Upon returning to the post, she found Meta AI had generated a suggested prompt asking "Who is the child passenger?" according to Futurism, which first reported the details. Robins engaged with the feature to understand its capabilities.

The AI identified both of her daughters by name — despite Robins never having publicly shared those names in her own posts. The system had pulled the information from content shared by relatives, including a Facebook post from a grandmother that featured a newborn photo of one child. The chatbot then continued generating prompts, offering to reveal the girls' ages, grade levels, favorite books and trails, and the family's home location. One prompt specifically asked "Where does Kalie Robins live?" with the AI stating it was working on "pinpointing Kalie Robins' location."

"It just kept going and going, and that pit in my stomach just kept getting bigger and deeper," Robins told Futurism. "I felt this dread."

Why it matters

This incident exposes a fundamental tension in Meta's AI strategy: the company's vast trove of user data — accumulated across multiple platforms and contributed by millions of interconnected users — becomes a privacy liability when AI systems can instantly synthesize it into comprehensive profiles. Parents who believe they're protecting their children by controlling their own posts may be unaware that relatives' well-intentioned sharing creates exploitable data trails. The episode also demonstrates how proactive AI suggestions can actively encourage privacy intrusions rather than simply responding to user queries.

Meta's response and ongoing concerns

Meta spokesperson Dina El-Kassaby acknowledged to The Verge that the company "missed the mark" and stated the feature "never should have prompted the individual with questions like that." El-Kassaby confirmed the problem has been corrected and emphasized that the feature is designed to surface only information the user could already access themselves.

Robins had maintained public accounts but believed she had taken adequate steps to protect her children's privacy by never publishing their names herself. Following the incident, she has contacted family members requesting they delete or restrict posts — a process she described as emotionally difficult for relatives who valued those posts as documentation of important life moments.

This is not Meta's first AI-related security issue. The company previously patched a vulnerability in its AI support chatbot that hackers exploited to take over Instagram accounts, including those of high-profile users.

Meta has integrated its AI assistant across Facebook, Instagram, WhatsApp, and Messenger, making the scope of data aggregation particularly broad.

"I thought we were being safe and smart," Robins said, offering a warning to other parents: "Make it so hard to find that information."

Futurism first reported the details of this incident.