A Claude sharing feature turned private-looking conversations into public webpages—and some were discovered by search engines.
Sharing an AI conversation with a friend or coworker can feel like sending a private document. In reality, the resulting link may function more like a page published on the open web.
Claude conversations are private by default. However, when a user selects the Share option, Anthropic creates a public snapshot that anyone with the link can view. Recent reports found that some of those shared conversations and Claude Artifacts—including documents, business plans and technical diagrams—had appeared in Google and Bing results.
According to Axios, a shared page must first be posted somewhere a search engine can discover it, such as a public forum or social network. Anthropic says it does not provide search engines with directories or sitemaps of shared conversations.
Still, WIRED reported that some pages lacked a standard “noindex” instruction commonly used to keep public URLs out of search results. Google listings later disappeared, although Bing continued to display shared pages at the time of WIRED’s reporting.
The distinction matters: this was not a breach exposing every Claude user’s private history. Only conversations intentionally turned into public links were affected. But users may not realize that “anyone with the link” can eventually become “anyone using a search engine.”
Anyone who has shared a Claude conversation can review active links under Settings, Privacy, Shared Chats. Anthropic’s privacy guide explains how to revoke access by changing a shared conversation from public to private.
The safest rule is simple: before sharing an AI chat, remove passwords, financial information, customer data and anything else you would not want appearing on a public webpage.
