Hundreds of thousands of conversations that users with Elon Musk’s Xai Chatbot Grok are easily accessible through Google search, Forbe reports.
When a Grok user clicks the “Share” button in a conversation with a chatbot, a unique URL is created that users can use to share their conversations via email, text, or social media. According to Forbes, these URLs are indexed by search engines such as Google, Bing, and Duckduckgo.
Users of Meta and Openai chatbots have recently been affected by a similar issue. Similar to these cases, the chats Grok leaked offer a glimpse into questions about the less valuable desires of users: how to hack Crypto Wallets. Dirty chat with explicit AI persona. Ask for instructions regarding the scalpel cooking.
Xai’s rules prohibit the use of bots to develop “promote critically harming human lives” or “biological, chemical, or weapons of mass destruction,” but that doesn’t stop users from asking Grok for help with such things anyway.
According to a conversation that Google can access, Grok gave users instructions on how to make fentanyl, handed out various suicide methods, tips on building bombs, and even had detailed plans for the assassination of Elon Musk.
Xai did not respond immediately to requests for comment. I also asked when Xai began indexing Grok conversations’ indexes.
Later last month, ChatGpt users sounded an alarm on Google that their chats were indexed. In a post with the quote from Musk with the word “Grok ftw,” Grok explained that “there is no such sharing feature” and “there is prioritizing.[s] privacy. “
TechCrunch Events
San Francisco
|
October 27th-29th, 2025
Do you have sensitive tips or confidential documents? We report on the internal mechanisms of the AI industry. From companies shaping their futures to those affected by their decisions. Contact Rebecca Bellan and Maxwell Zeff at maxwell.zeff@techcrunch.com. For secure communication, please contact us via the signals @rebeccabellan.491 and @mzeff.88.
Source link
