Meta Changes AI Prompts After Family Privacy Backlash

Meta says it has changed an AI suggestion feature after a user showed the chatbot assembling detailed information about her young children from years of activity across Facebook and Instagram. The episode turned an automated prompt beneath a family video into a sharp warning about how easily scattered public details can be collected into one profile.

Utah-based creator Kalie Robins said the experience began after she cross-posted a video of herself singing in a car with one daughter. Meta AI displayed a suggested question asking who the child passenger was. When Robins opened it, the system identified both daughters and offered further prompts about their ages, interests and the family’s location.

Old posts became a ready-made dossier

Robins told Futurism that the chatbot appeared to draw not only from her own public accounts but also from posts shared by relatives over several years. Screenshots described in the reporting showed the system referencing birth announcements, family photographs and other routine updates that had never been collected in one obvious place.

The concern was not that each fragment was necessarily secret. It was the speed and confidence with which the assistant assembled those fragments, including information about children, and then encouraged the user to keep asking for more. Robins has since asked relatives to remove or hide older family posts, a process she described as emotionally difficult because those accounts had also served as shared scrapbooks.

Meta says the prompt missed the mark

A Meta spokesperson told both The Verge and Futurism that the feature was intended to help people learn more about posts and topics that interest them. The company said generated answers use information the user can already access, including public web material and the user’s own content.

Meta nevertheless acknowledged that the system should not have suggested questions of this kind and said the underlying issue had been fixed. The change addresses the prompt behavior described in this case, but the incident leaves a broader question for families using social platforms: information shared innocently over many years can become far more revealing when an AI tool gathers it instantly. For users, privacy now depends not only on what one account posts, but also on the digital trail created by an extended network of relatives and friends.