Meta says it has patched a flaw in its Meta AI system after an Instagram creator went public accusing the tool of stitching together years of her old posts to expose private details about her family, including her children.
Kalie Robins, who posts under the handle kontheinside, posted a video in early September describing what happened after she shared a clip of herself singing in a car with her daughter in the backseat. Beneath that post, Meta AI offered up suggested prompts including “Who’s the child passenger?” and “Where does Kalie Robins live?”
Robins said clicking those prompts sent the AI combing back through her account history to piece together private information, including her location. She said the system also drew on other accounts, including her mother’s, and surfaced material from a photo she believed she had deleted years earlier.
“I did not understand posting separate innocent pieces of our lives over years could mean an AI could sit there and assemble all those breadcrumbs into information about my minor children in seconds,” Robins said in the video.
The video took off, pulling in thousands of comments and more than 308,000 likes. Some parents responding to the post began urging others to stop putting images of their kids online altogether.
A Meta spokesperson told CNET the problem has been “fixed” and that Meta AI will no longer suggest personal, invasive prompts on posts. The company said the feature is only supposed to help people locate information they already have access to, not generate new profiles of strangers’ private lives.
Meta pushed back on one detail of Robins’ account. The company told CNET the AI never actually accessed a photo she’d deleted years earlier — instead, it said, that photo had only just been deleted before she made her video, and a temporary bug kept it visible a little longer than it should have been.
Robins isn’t buying the explanation. In a follow-up post on September 10, she demanded specifics: “What exactly was fixed? What safeguard failed in the first place? And what protections are now in place when Meta AI encounters information involving a child?”
She went further in an email to CNET, rejecting Meta’s attempt to reduce the incident to a single photo. “The issue was never just one photo, it was the AI compiling information about my kids from multiple sources,” she said. She added that Meta has never contacted her directly or apologized.
For a company that has spent years insisting it takes user privacy seriously, the episode is another reminder that the burden of catching these failures still falls on the users themselves — and that the company’s response, even after going viral, has been a quiet software patch rather than direct outreach to the family affected.
