Users reported that a trove of apparently private Anthropic Claude conversations appeared in Google search results after using Claude’s “share” feature. The exposed items reportedly included sensitive information such as cryptocurrency wallet keys and personal details like names and addresses, according to reports surfaced by third-party investigators. Anthropic said it provides users control over sharing and that it does not share chat directories or sitemaps with search engines; it framed the issue as shared conversation links becoming publicly accessible by the user’s choice, and it said it has fixed the discoverability method. However, many previously exposed links remained reachable via the shared URLs while the problem was active. For higher education IT teams and research units deploying consumer AI tools, the incident is a direct reminder to treat “share” links and public indexing as potential privacy and compliance risks, particularly when students or researchers paste institutional or personal data into chat workflows.