Hundreds of private conversations with Anthropic’s Claude AI chatbot were publicly accessible through search engines, exposing personal details, work-related information and even transcripts of healthcare discussions.
The chats were discovered by users on Reddit, who found that links to conversations – shared via Claude’s “share” feature – had been indexed by Google and appeared in search results. Over 200 conversations across at least 25 pages of results were visible, some from as recently as weeks ago.
“Hundreds of private conversations with Anthropic's Claude AI were publicly accessible via search engines for weeks.”
Among the exposed logs was an exchange from April in which a user asked Claude to draft an unpublished blog post about cloud security, revealing details of a corporate project. Another from last month saw a user ask how to “become Nine-tailed fox?” – a request to transform from human to creature – to which Claude responded with an AI-generated image claiming the user had been given “fully functional fox powers!”.
Other conversations included users seeking help with CVs, disclosing names, contact information and work history. Some appeared to involve proprietary research in healthcare, including transcripts of private conversations.
A spokeswoman for Anthropic said users maintained control over sharing: links were “not guessable or discoverable unless people choose to share them themselves”. She added that shared content “may be archived by third-party services”. Claude’s share option warns that “anyone with the link” can view the conversation, but does not explicitly mention search engine indexing.
The search availability of the chat logs was removed over the weekend, but by then many had been saved and shared widely. The incident mirrors a similar problem with OpenAI’s ChatGPT last year, which prompted the company to change how easily such logs could be accessed.