Google search results exposed publicly shared Claude AI conversations.

Concerns about AI privacy resurfaced after users discovered that some Claude AI shared conversations were appearing in Google Search results. A Reddit post showed that searching “site:claude.ai/share” returned publicly accessible chat pages, including discussions containing personal details and sensitive subjects.

The discovery quickly sparked criticism, with some users questioning whether Anthropic had done enough to prevent shared conversations from becoming searchable.

Anthropic Says It Was Not a Data Breach

Anthropic denied allegations that Claude secretly fed conversations to search engines. The company argued that its sharing feature only created a publicly accessible link if the user wanted to share a conversation. Anthropic stated that these links were indexed by Google’s crawlers because users shared them on public forums.

That distinction is important because it means the incident was not caused by hackers or unauthorized access. Instead, it highlights how public links can spread beyond their intended audience once they are shared online.

A Reminder About AI Sharing Features

The incident mirrors a similar situation previously seen with ChatGPT, showing that AI chat-sharing tools can create privacy risks if users assume a shared link remains private.

As AI assistants increasingly include collaboration and sharing features, companies face growing pressure to make the privacy implications clearer before users publish conversations containing personal or confidential information.

For the average user, the big take-home from this episode is that if a chat link is public, treat it like any other website that could potentially get indexed or shared more broadly. The episode will no doubt lead to changes in how AI firms present public-facing sharing options and warn users of the risks of making chats discoverable on the web.