Published Claude AI chats indexed by Google: check if your conversations are exposed

Over the weekend, Reddit users discovered that a simple search query site:claude.ai/share revealed published Claude AI chats — ranging from adult fan fiction to discussions of news, creation of fake book reviews, and conversations about programming. Anthropic responded quickly: the Google dork stopped working shortly after publication, and the chat lists disappeared from Google. However, many of the links are still active, and according to ZDNet, their content could have financial or legal implications. Some users claim to have found chats containing cryptocurrency wallets, though this is unconfirmed. Unlike the situation in June, when Anthropic revoked access to Fable 5 and Mythos 5 at the White House's request, this time the leak was unintentional: chats created via the Share feature were not protected from being indexed by search engines. Personal chats remain private by default and were not disclosed. Users are advised to check their settings under Settings > Privacy > Shared Chats and disable active links for publication. ZDNet has reached out to Anthropic for comment.