OpenAI’s short-lived sharing feature turned personal AI chats into searchable, OSINT-ready data.

💡 Not a Medium member? You can still read this article in full with [Click here]

Thousands of private conversations with OpenAI’s ChatGPT recently became searchable on Google — not through a hack, but through a design flaw masquerading as a “sharing feature.” In doing so, OpenAI inadvertently gave the world a glimpse into how AI-driven conversations can be weaponized as open-source intelligence (OSINT) at scale.

In July 2025, OpenAI quietly rolled out an experimental toggle labeled “Make this chat discoverable” during the chat-sharing process. If a user clicked both “Share” and that checkbox, their conversation was indexed by search engines like Google and Bing.

While the feature was opt-in, many users were shocked to learn their personal conversations were turning up in search results, sometimes alongside identifiable details. From mental health confessions and workplace grievances to drug use, sex lives, and job applications, over 4,500 ChatGPT chats were indexed, according to a Fast Company investigation. Privacy experts now believe that figure was just the tip of the iceberg.