Users Believed It Was ''Private,'' But the Platform Interpreted It as ''Public Feed''
Meta AI app's "Discover" feed is publicly exposing users' personal conversations, images, and voice messages — medical diagnoses, legal inquiries, relationship concerns, official documents — appearing alongside real names and Instagram profiles. The core problem: when users share AI conversations in the Discover feed, there is insufficient warning or reconfirmation process, leading many users to unknowingly share sensitive information. Users typically discover their information has been publicized through warnings from strangers or friends. Meta AI (based on LLaMA 3, Emu, Code LLaMA models) is integrated into Instagram, Messenger, WhatsApp, and Facebook, with the Discover feed designed to share AI-generated content. Design flaw analysis: the sharing flow embeds "share to Discover" as a default or prominently displayed option after AI conversations, exploiting the "I just finished a helpful conversation" positive emotional state to generate content shares. Privacy experts note Meta AI's sharing mechanisms are less transparent than competitors', with serious UX design defects. EU and civil society organizations demand temporary suspension of Discover feed until privacy protection measures are strengthened. GDPR context: Meta claims "legitimate interest" as legal basis for data processing, but using it to justify AI training on sensitive personal conversations shared via inadvertent UI design is legally contested. The broader pattern: AI companies are discovering that making AI assistants also content generators creates privacy conflicts — users seek private advisors but platforms incentivize public sharing, creating systematic exposure of information users would never intentionally make public.
