Meta Confirms Fix for Meta AI Privacy Recommendation Flaw

Meta has confirmed that it has fixed a flaw in its AI assistant, Meta AI, which previously prompted users with inappropriate questions about personal privacy and could compile profiles of individuals, including children, from information accessible to users.

The controversy stems from an Instagram video posted by parenting blogger Kalie Robins on September 2. The video showed Robins singing in her car with her daughter in the back seat. Meta AI automatically generated a series of questions below the video, including “Who is the child in the back?” and “Where does Kalie Robins live?” When users clicked these questions, Meta AI extracted information from her previous posts, piecing together personal details scattered across different posts over the years, such as her location and her mother's account.

As the issue gained traction, Robins's Instagram comments saw extensive debate, with some parents urging others to limit sharing children's photos online to reduce AI-related privacy risks. A Meta spokesperson told CNET that the company has “fixed” the issue and that Meta AI will no longer recommend inappropriate personal privacy questions based on user content. Meta acknowledged that the feature had not met expectations and emphasized that the design purpose of Meta AI is to help users get more from information they already have access to.

In a follow-up post on September 10, Robins said she was glad Meta acknowledged the issue but still had questions, including what exactly was fixed, which safety measure initially failed, and what protections are now in place when Meta AI encounters information about children. She argued that the core issue is not whether the feature is enabled or disabled, but that Meta AI is designed to collect and organize private user information from multiple sources. As of now, Meta has not directly contacted or apologized to her.