Overview

• Users of Meta AI have unintentionally published private text, audio and image conversations to the Discover feed, linking posts to their Facebook and Instagram identities.
• Since June, Meta has introduced a pop-up alert and an additional confirmation step to warn users and slow down the sharing process, leading to a sharp drop in text-based posts on the Discover feed.
• Privacy experts at the Electronic Privacy Information Center and Mozilla Foundation say the app's default settings and unclear interface still risk exposing sensitive medical, legal and financial details without explicit user understanding.
• Regulators in the US and UK are scrutinizing Meta's AI data handling practices under emerging privacy and AI oversight guidelines.
• Users can retroactively remove shared prompts by adjusting privacy settings in the app under Data & Privacy, but experts warn post-publication deletion cannot undo potential misuse or screenshots.