Google Revealed Private ChatGPT Conversations in Search Results

The Public Could Search Thousands of Shared AI Chats, Even the Most Private Ones

For a while, ChatGPT seemed to be the safe haven people needed, where they could vent to AI about trauma, breakups, or difficulties in life before possibly sending a link to a friend. However, the “Share” button? It had different ideas.

Google began indexing ChatGPT’s shared discussions, allowing anyone with the appropriate keywords to browse private exchanges. Anyone who knew how to look may see thousands of user-submitted chats that appeared in search results, ranging from confessions of addiction and abuse to worries about being watched.

When users clicked ChatGPT’s “Share” button, the problem started. That button generates a unique URL, which users can save for later or distribute to a friend via text message or WhatsApp. Many were unaware that the talk could appear in Google results if the “Make this chat discoverable” option was chosen.

Approximately 4,500 chats were reportedly captured in this digital dragnet. Some revealed very intimate information—true tales of relationship difficulties, employment stress, mental health struggles, and more—while others were generic or innocuous.
People frequently revealed so many personal details that it would be easy to figure out who they were talking about—or who they were themselves—even if OpenAI didn’t include names.

Since then, OpenAI has eliminated the ability to make shared chats “discoverable.” In order to clean up the digital spill, they are also collaborating with Google to remove indexed links.

This is a warning to all users of ChatGPT’s Share feature. Your link might still be out there if it was public. The wisest move? Remember that nothing on the internet is really private, so go through your shared messages and remove anything you don’t want to be seen publicly. Not even the conversations you have with a robot.