Claude's Shared Chats and Artifacts Were Turning Up on Google

Some Claude users who shared conversations or Artifacts using Anthropic's "share chat" link feature found those links indexed and searchable on Google. The feature is designed to let anyone with the assigned URL view a conversation or project — but that appears to have extended, in some cases, to search engines crawling and surfacing the same links.

What happened

Anthropic's Claude includes a share-chat option that generates a public URL for a conversation or Artifact, intended for anyone holding that link to view it. According to the report, some of those URLs ended up indexed by Google, meaning conversations or projects that users may have assumed were only accessible to people they directly shared the link with became discoverable through a search engine instead.

[VERIFY]: The exact scope of the exposure — how many links were indexed, over what time period, and whether Anthropic has issued a fix or official statement — is not detailed in the available summary and should be confirmed before further coverage.

Why it matters

For consumers: anyone who shared a Claude conversation or Artifact assuming only their direct recipient could see it should check whether that content included sensitive information — personal details, unpublished work, or anything not meant for public view.

For developers and builders: the incident is a reminder that "shareable link" and "private by default" are not the same guarantee, especially once a URL becomes crawlable. Teams building similar link-sharing features in their own products should double-check whether those links are disallowed in robots.txt or otherwise blocked from indexing.

For business decision-makers: enterprises using Claude for internal work should review whether any shared links touched confidential or regulated data, and watch for Anthropic's guidance on rotating or revoking exposed links.

What's unknown

Anthropic's official response, the root cause of the indexing, and whether the share-link behavior has already been changed are all [VERIFY] pending further reporting or a statement from the company.

What to watch

• Whether Anthropic publishes a statement or postmortem on the exposure
• Any changes to default privacy settings or indexing controls for shared Claude links
• Whether other AI chat products with similar sharing features face the same issue
• Guidance from Anthropic on auditing or revoking previously shared links

Source: TechCrunch