OpenAI Google In mid-2024, security researchers discovered that thousands of ChatGPT conversations shared via the "Share" feature were fully indexed by Google. That therapy session you had with the AI? Now on page 1 of search results. The Discovery Researchers found over 100,000 conversations archived by the Internet Archive. The initial count of 4,500+ indexed links was just the tip of the iceberg. What Was Exposed The leaked conversations included: Personal resumes with contact details Mental health discussions and therapy-like conversations Business strategies and proprietary plans API keys and developer credentials Client names and internal research Deeply personal emotional content Many users didn't realize checking "Make this chat discoverable" meant "publish to the entire internet." OpenAI's Response August 1, 2024: Removed the "discoverable" feature entirely Added and tags Working with Google to de-index existing URLs Called it a "short-lived experiment" But the damage was done. Internet Archive saved copies. Cached pages persist. The Technical Failure ChatGPT Enterprise customers were exempt—their shared links were never indexed. But regular users? No protection. No restrictions initially No clear warning that links would be searchable Deleting a chat does NOT delete the shared link What to Do Next Check Your Exposure: Go to ChatGPT Settings → Data Controls → Shared Links Review all shared conversations Delete individual links or use "Delete All" Search Google: Request Removal: Use Google's Remove Outdated Content tool if you find exposed chats. The Lesson: Every AI conversation is one button click away from public exposure.