Thousands of Claude Conversations with Sensitive Data Found Publicly Searchable on Google

Summary

An unconfirmed number of Claude conversations and Artifacts were discovered to be publicly searchable on Google over the weekend after Reddit users identified that specific search operators (like 'site:claude.ai/share') could surface shared chats. The exposure was traced to Claude's "share chat" feature, which creates publicly accessible links intended for small-group sharing but ended up indexed by search engines, making them discoverable to anyone.

The exposed content included highly sensitive information: health records, medical reports with patient names, clinical trial results, company documents marked for internal use only, employee reviews with personal information, and phone numbers of children. Some Artifacts contained code, work notes, and explicit sexual content—including at least one conversation labeled "shared by Anthropic" that violated the company's own policy prohibiting the generation of such material.

Anthropicfocused its response on user responsibility, arguing that shared links only appear in search results when users post them on publicly accessible platforms like forums or social media. The company stated it does not provide chat directories or sitemaps to search engines. By Monday afternoon, Google searches using the method that exposed the conversations returned no results, suggesting the issue had been remediated. This marks the second known incident of Claude conversations being indexed by search engines; approximately 600 were indexed in a similar 2025 incident.

• The exposure was remediated by Monday afternoon, but the total number of conversations indexed and duration of exposure remain unconfirmed