Anthropic's Claude AI shared chats appear in Google searches, raising privacy concerns

Anthropic made statements saying it does not provide search engines with directories or sitemaps containing user conversations, emphasizing that only chats intentionally shared through public links can become accessible on the open web. The company maintained that Claude conversations remain private by default unless users choose to generate a shareable URL.

Google, meanwhile, reiterated that website operators are responsible for controlling whether publicly accessible pages are indexed by search engines. Publicly shared Claude pages were found to lack the 'noindex' metadata commonly used to prevent indexing, allowing some conversations to appear in search results.

Reportedly, several indexed pages may contain sensitive material, including internal planning documents, legal discussions, and personal conversations. Although there is no indication that Claude exposed private chats without user action, the incident illustrates how easily public sharing features can be mistaken for limited-access collaboration tools. Similar concerns have surfaced previously around other AI platforms that allow users to share conversations via publicly accessible links.

All of this leads to a relatively straightforward immediate solution for users: a shared conversation link should be treated like any other public webpage unless explicit protections are CLEARLY in place. The incident also serves as another reminder that AI platforms should take a closer look at making the distinction between "shared" and "publicly searchable" clearer to the average user, particularly as generative AI tools become increasingly integrated into professional and collaborative workflows.