Claude's share chat links may have been indexed by Google

Claude’s “share chat” feature, which generates URLs allowing anyone with the link to view a conversation or project, has been identified as the source of a recent privacy concern. Users reported that the links could be accessed without additional authentication, raising the possibility that sensitive information shared through the AI platform could be inadvertently exposed to unintended parties.

Anthropic, the developer of Claude, confirmed that the issue stemmed from the default settings of the sharing mechanism and is working to implement stricter access controls and clearer user warnings. The company has advised users to review existing shared links and to revoke any that are no longer needed while it rolls out the security enhancements. The incident underscores the importance of robust permission structures for AI collaboration tools.