Shared Claude conversations turned up in Google search results

Private Claude chats showed up in Google search results, and the exposed material was not trivial. On July 27, 2026, TechCrunch and Futurism reported that shared Claude conversations and Artifacts, published via the chatbot’s share button, were indexed by search engines and searchable by anyone.

What was exposed

The reviewed materials included a patient’s medical report, clinical trial results with patient names, contact details of primary-school children, and internal company documents. These were conversations users had shared by link, presumably with a specific colleague or friend in mind, not with the open web.

Anthropic’s response: working as intended

Anthropic’s position is that the share feature did exactly what it was designed to do. A share link makes the page public, and public pages can be indexed by search engines. Technically, that is accurate.

The catch is what users understood. There was no warning that a link meant for a colleague could end up in Google. The mental model most people have of a share link, a semi-private URL that only the recipient will see, does not match what the feature actually does.

Not the first wave

This has happened before. Forbes reported hundreds of indexed Claude chats last year, which Anthropic then scrubbed from the web. The recurrence suggests the underlying design tension, convenience of sharing versus clarity about publicity, has not been resolved.

What to do about it

The practical takeaway is simple: if you ever sent anyone a Claude chat link, that page may be public right now. Checking your share history and revoking links you no longer need is a five-minute audit worth doing, especially if any of those chats touched health, legal, or company-internal topics.

The broader lesson applies to every AI tool with a share button. Until the industry converges on clearer warnings, the safe assumption is that a shared AI conversation is a published one.