Elon Musk’s AI venture, xAI, has accidentally left hundreds of thousands of Grok chatbot conversations publicly accessible on Google, sparking major privacy concerns. Reports Technology News
The issue stems from Grok’s “share” feature. When users opted to share a chat, the system generated a unique URL intended for private distribution via email, text, or other channels. However, these links were also indexed by search engines such as Google, Bing, and DuckDuckGo—effectively publishing private conversations online without user awareness.
According to a Forbes investigation, more than 370,000 Grok chats were visible on Google. The leaked pages ranged from simple requests, like drafting social media posts, to highly dangerous content. Some transcripts revealed Grok offering guidance on illegal activities, including drug manufacturing, explosives building, malware creation, and even plotting an assassination attempt on Musk himself.
Many conversations also exposed sensitive personal information. Users had sought medical and psychological advice, shared personal details, passwords, and names, and even uploaded documents such as spreadsheets and images—all of which became publicly available through the shared links.
