Meta fixes Instagram AI privacy flaw after personal questions surfaced

Meta has announced it has corrected an artificial intelligence feature that generated inappropriate personal questions about Instagram users and their families. The fix follows a viral demonstration by parenting creator Kalie Robins, who showed how Meta AI suggested prompts such as “Who’s the child passenger?” and “Where does Kalie Robins live?” beneath her video posts.

Robins demonstrated that by selecting these prompts, the AI tool could rapidly aggregate fragmented information from years of social media posts, including content from other profiles like her mother’s account, to assemble specific details about her minor children and location. The incident highlighted concerns regarding how AI can combine publicly available data points more quickly than a human user.

Meta stated that the feature was intended to help users find information they already have access to, but acknowledged the prompts were inappropriate. While the company claims the issue is resolved and will no longer generate such questions, Robins has raised further questions regarding the specific safety measures implemented and how the AI handles information related to minors.