Meta Removes AI Prompts After Instagram Privacy Scare

Meta removed problematic AI-generated prompts on Instagram after a lifestyle blogger showed the tool suggesting questions that exposed personal details about her child and home.

The video came from lifestyle blogger Kiley Robbins, who posts as @kontheinside on Instagram. Robbins showed that AI prompts under an old video of her singing in a car with her daughter included questions such as "Who is the child with?" and "Where does Kiley Robbins live?"

Robbins said clicking those system-generated questions caused the AI to aggregate information from her older posts and other accounts, including her mother’s, and to assemble private details such as her residence. She also said the tool surfaced a photo that had been deleted years earlier.

The video quickly drew thousands of comments and received more than 308,000 likes.

A Meta spokesperson said the issue has been fixed and that the tool will no longer suggest inappropriate personal questions on users' posts. Meta acknowledged the feature did not meet expectations and said the AI is intended only to help users access information they already have.

Robbins said Meta’s acknowledgement was welcome but left key questions unanswered, including what exactly failed, which protective measures were bypassed, and what protocols now govern AI handling of children’s data.

Meta said it did not retain or access the deleted photo and that the image had been removed just before production of the video; the company said a temporary bug made it briefly available. Meta also noted the AI can still use information already accessible to it, such as public text, web images, or users' older posts.

Robbins says the core problem is broader than a single image: she warns the AI can compile data about children from multiple sources. Meta has not directly contacted Robbins or issued a personal apology.