Meta AI Asked Disturbing Questions About Children, Mom Claims

A Utah mother has described a concerning experience with Meta's AI tool, alleging it asked invasive questions about her children and uncovered private photos that had been deleted for years. The incident began after Kalie Robins posted an Instagram video of herself and one of her daughters on September 1.

The following day, after revisiting the post due to a copyright notification from Meta, Robins encountered an AI-generated suggestion: "Who’s the child passenger?" Robins expressed her alarm at the nature of the question, particularly given it concerned a minor. Upon clicking the suggestion, she reported that the AI began compiling information about both of her daughters, including one not present in the video.

"It starts pulling completely separate information for each of my kids - their names, birth information, pictures and videos of them," Robins stated, according to reports. The data allegedly included posts and images from various social media accounts, even sourcing a newborn photo of her child from Robins' mother. The AI then reportedly suggested another query: "Where does Kalie Robins live?"

Robins shared her frustration, stating that the AI was piecing together information from unrelated sources across her digital history to reveal her location. She argued that while users accept risks when posting online, AI should not facilitate the tracking of families and children.

Robins declared she was ending her use of Meta platforms and urged friends and family to remove any identifiable images of her children from their accounts, as the AI could retrieve such information from any user's profile accessible to it. She posted a warning on Instagram Reels, advising others to save memories privately and remove children's images from public pages.

Meta acknowledged the incident, stating that the AI feature is designed to help users find information about topics of interest using data they already have access to. However, the company admitted its questions "missed the mark" in this instance. A Meta spokesperson told Futurism, "we’ve fixed that issue" regarding the inappropriate prompts.

Despite Meta's assurances, Robins questioned the specifics of the fix, asking what safeguards failed and what protections are now in place for children's data. The incident occurs amid broader concerns about the advancement and unchecked development of artificial intelligence, with some former employees and researchers warning of potential risks, including claims that AI could pose existential threats or contribute to suicides.