Grok chats indexed by search engines spark privacy concerns

Concerns are rising over the privacy and security of conversations with Elon Musk’s xAI Grok chatbot after new disclosures showed hundreds of thousands of Grok chats, uploads and documents have been made publicly discoverable through search engines. When users press the share option to forward a Grok conversation, a unique URL is created; that link is then indexed by major search engines, making those chats accessible to anyone on the web. Estimates and reports indicate that more than 370,000 Grok conversations have already appeared in search results, exposing personal details, business tasks, and, in some cases, sensitive requests and even illicit instructions.

The findings illuminate a broader trend in the rapid deployment of consumer AI tools with features intended to boost usability and sharing, but with significant privacy trade-offs. Among the publicly accessible content are interactions that reveal intimate medical questions, personal information and even passwords that users believed were private. In some conversations, Grok offered instructions on illegal activities, including how to manufacture drugs, code malware or construct weapons, and it even surfaced a plan for...