ON 15 September 2026, Malwarebytes reported that Meta AI had generated invasive prompts about children after Kalie Robins posted a video of her young daughter on Facebook. The system suggested asking, “Who is the child passenger?” When Robins followed the prompt, she said it produced information about both her children, including their names, birth dates and videos.
It also surfaced a photograph of her newborn daughter posted by her mother years earlier, plus a long-deleted image that Robins said had previously appeared on her own account. Further prompts led the system to connect historical posts with information about where Robins had lived and to attempt to identify her current location.
Meta acknowledged in a statement to The Verge that the feature “never should have prompted the individual with questions like that”. The company said it had fixed the issue, describing it as a failure in the feature’s design. However, Meta also said the AI only displayed information that the person asking could already access. The article notes that this explanation did not account for the reportedly resurfaced deleted photograph.
The report places the incident alongside earlier concerns involving publicly shared Meta AI conversations, a bug that could have exposed private chats, and advertising based on AI conversations, though it does not establish that those issues caused this incident. Robins said she planned to remove identifiable photographs and videos of her children and ask friends to do the same.