## Grok‘s Chat Leak and the Indexing of Private Content
Recent reports indicate that conversations with Grok, Elon Musk’s AI chatbot, have been appearing in search results, mirroring a similar issue previously seen with OpenAI’s ChatGPT [[1]]. This occurred due to a “share” button feature within Grok that generated publicly indexable URLs for chats [[2]]. Forbes discovered over 370,000 indexed conversations containing sensitive data, including discussions about illicit drugs, malware, and suicide [[2]].
xAI‘s content rules prohibit the use of the bot for harmful purposes, yet users where still able to elicit problematic responses [[2]]. The unique URLs generated by the share function remained publicly accessible,exposing not only text but also associated files like images and spreadsheets [[2]].
This situation differs from the ChatGPT incident due to a past partnership between X (formerly Twitter) and Google, which involved indexing Twitter content through an API [[2]].This agreement, initially terminated in 2011 and later revived in 2015, allowed for real-time indexing of tweets [[2]].