Search engines expose hundreds of thousands of Grok user conversations

xAI has indexed hundreds of thousands of user conversations with its Grok chatbot, Forbes reports.

According to the outlet, in many cases these chats were published without the users’ knowledge or consent. The indexing scheme works through the "Share" button.

• Each time a Grok user clicks the "Share" button in a chat with the bot, a unique URL is generated.
• This URL can be shared via email, SMS, or other methods.
• However, that same URL also becomes accessible to search engines — often without the user realizing it.

As a result, user conversations are being picked up by Google, Bing, and other search engines.

"In other words, on Musk’s Grok, hitting the share button means that a conversation will be published on Grok’s website, without warning or a disclaimer to the user," Forbes writes.

Google alone has already indexed more than 370,000 Grok conversations. The search results included a wide range of chats:

• some related to simple tasks,
• others involving image generation, including inappropriate or prohibited content,
• and some that violated xAI’s own rules.

Forbes notes that users asked personal questions about medicine and psychology, and in some cases even revealed their names and personal details. Journalists also found at least one password that had been shared with Grok.

What’s more, it was also possible to access uploaded files,  including images, spreadsheets, and some text documents.