Meta’s new AI Discovery feed has been found to publicly display users’ seemingly private chats, including deeply personal conversations, images, and audio clips. The issue, which has surfaced just months after the launch of Meta’s standalone AI app, has ignited a firestorm of criticism and raised urgent questions about user safety and data protection on one of the world’s largest technology platforms.
The controversy centers on the Meta AI app’s “Discover” feed; a feature designed to showcase AI-generated content and conversations. However, instead of merely highlighting curated or intentionally shared exchanges, the feed has been inundated with highly sensitive user data. Reports confirm that medical inquiries, legal questions, relationship doubts, and even employment documents are now visible to anyone browsing the app, often tied directly to users’ real names and public Instagram profiles.
At the heart of the problem is the app’s “share” button. After interacting with Meta AI, users are prompted to share their conversation. While Meta claims this is an opt-in process, the interface reportedly provides minimal warnings and no robust double-check mechanism. This has led many users to unwittingly publish private conversations, believing they were simply saving or bookmarking their chats rather than broadcasting them to the world.
The scope of exposed information is staggering. Posts in the Discovery feed have included:
Medical diagnoses and health-related questions
Legal advice and immigration inquiries
Personal confessions, relationship issues, and mental health crises
Voice recordings, photos, and even official documents like rental agreements and employment letters
Many users only realized their sensitive data was public after being contacted by strangers or warned by friends, amplifying the sense of violation and loss of trust in the platform.
Privacy experts and user advocates point to a serious user experience (UX) failure. Unlike rival AI chatbots such as ChatGPT or Google Gemini, which offer clear private links and explicit sharing controls, Meta AI’s sharing process defaults to public with only a small, easy-to-miss disclaimer. The lack of prominent warnings or a memory-free mode means users are left vulnerable to accidental oversharing.
Compounding the issue, users who log into Meta AI with their Instagram accounts may find their chatbot activity publicly linked to their social profiles, further increasing the risk of identity exposure and harassment.
The revelations have triggered widespread outrage. Privacy advocates and European Union regulators are now demanding stricter controls or even a temporary suspension of the Discovery feed until Meta addresses the glaring flaws in its privacy design. Critics argue that Meta’s push to integrate AI across its platforms has outpaced its ability to protect users, leaving millions at risk.