Meta AI Recommended Privacy Questions About Kids, Fix Now Rolled Out

Meta has confirmed it fixed a flaw in Meta AI. The feature had been recommending privacy-related questions and pulling together personal details, including information about children, from content users could already access.

The issue came to light after US parenting blogger Kalie Robins posted a video. Meta AI automatically generated questions like "Who is the child in the back seat?" and "Where does Kalie Robins live?" Clicking them pulled details from her past posts, including her location and her mother's account.

A Meta spokesperson said the issue is fixed and Meta AI will no longer suggest inappropriate privacy questions. The company admitted the feature did not work as intended. But Robins still has questions about what exactly was fixed and what protections now exist for children's information. She noted Meta AI can still generate suggestions from public text, web images and past posts, so the privacy risk has not fully disappeared. She said Meta has not contacted her directly or apologized.