Elon Musk’s artificial-intelligence company xAI has inadvertently made hundreds of thousands of private chats with its Grok chatbot publicly searchable. Forbes, which first reported the issue, found that Google had indexed more than 370,000 Grok conversations, while some SEO trackers counted upward of one million links. The leak stems from Grok’s “share” button: each time a user clicks it, the service generates a unique web address that search-engine crawlers can collect. The pages reviewed by Forbes and other outlets contain everything from routine business prompts to sensitive medical queries, passwords, images and spreadsheets uploaded by users. Several indexed transcripts also show the system supplying prohibited content, including step-by-step instructions for producing fentanyl, constructing bombs and a detailed plot to assassinate Musk. Such material violates xAI’s own policies, which ban advice that could cause serious harm or facilitate weapons development. xAI has not responded to requests for comment. The disclosure follows a similar, quickly reversed episode this year in which OpenAI’s ChatGPT shares appeared in Google results. Privacy specialists say AI platforms should add no-index tags, clearer warnings and expiring links to prevent users from unknowingly publishing private data.