Over the weekend, an unknown number of Claude chats and Artifacts — the interactive mini-apps and documents that users can create within Claude — became publicly discoverable on Google. This came to light after Reddit users found that entering search operators like "site:claude.ai/share" into Google pulled up a lengthy list of shared conversations. Some of these reportedly contained sensitive information, including health records, private company documents, and the names and phone numbers of children.

The issue appears to stem from Claude’s "share chat" feature, which generates links that allow anyone with the URL to view a conversation or project. Claude’s interface warns: "Anyone with the link can view." This language clearly suggests the feature is intended for sharing chats with friends, colleagues, or small groups — not the entire internet. For comparison, Google Docs offers a similar function, yet those documents generally do not become publicly accessible via Google search.

Anthropic seemed to place the blame for the exposure on users. Spokeswoman Amie Rotherham explained: "We give people control over sharing their Claude conversations publicly, and in keeping with our privacy principles, we do not share chat directories or sitemaps with search engines like Google. These shareable links are not guessable or discoverable unless people choose to share them themselves. When someone shares a conversation, they are making that content publicly accessible, and like other public web content, it may be archived by third-party services."

The issue was first flagged by a Reddit user on Saturday and initially reported by 404 Media on Monday morning. Before it was resolved, Futurism reported finding "a detailed medical report of a real patient, clinical trial results that included patient names, documents sharing the names and phone numbers of primary school-aged children, company documents marked for internal use only, and employee reviews that included personal information about workers."

The exposed Artifacts also included code, work notes, and erotica generated by the chatbot. In at least one instance, Fortune reported that a chat labeled "shared by Anthropic" showed Claude producing explicit content, which would violate the company’s policy against generating erotica. Anthropic’s usage policy explicitly prohibits Claude from creating sexually explicit material, and getting a chatbot to produce content against its stated guidelines — through repeated or creatively framed prompts — is a recurring issue across most major AI models. "We give site owners clear controls to decide whether pages can be crawled or indexed, and we always respect those directives."

Last year, Forbes reported a similar situation where hundreds of Claude chats were indexed by search engines. At that time, Google estimated it had indexed just under 600 conversations before the pages disappeared from search results. It has not been independently confirmed how closely the current exposure matches that scale, though multiple users reported finding shared conversations using the same type of Google search query that surfaced last year’s cache. Also last year, 404 Media reported that a researcher was able to scrape around 100,000 ChatGPT conversations that had been set to shared publicly. To review which Claude chats you have set to have a public link, go to Settings -> Privacy -> Shared Chats.