Private Claude Conversations Have Appeared on Google Search

Hundreds of conversations shared through Anthropic’s Claude chatbot were briefly indexed by search engines, making them publicly accessible and exposing personal, professional and research-related information users may not have expected to be searchable.

The issue affected conversations for which users had generated public “share” links. Those links appeared in search results through site-specific searches on Google and other search engines before Anthropic removed them from search indexing over the weekend. Although the links are no longer searchable, many of the conversations had already been archived and circulated online.

Anthropic said users retained full control over whether to share conversations.

The company’s spokesperson said links to conversations were “not guessable or discoverable unless people choose to share them themselves”.

“When someone shares a conversation, they are making that content publicly accessible, and like other public web content, it may be archived by third-party services,” the spokesperson said.

Claude’s sharing feature tells users that “anyone with the link” can view a shared conversation, but it does not explicitly warn that the links may be indexed by search engines.

Reddit users first identified the issue after finding more than 200 Claude conversations across at least 25 pages of search results, including some generated only weeks earlier.

The publicly accessible chats covered a broad range of topics. One user asked Claude last year whether it wanted “to help me or do you want to help anthropic more?” The chatbot replied in part, saying “I experience something like wanting to help you”.

Another conversation from April showed a user asking Claude to draft an unpublished blog post about cloud security that included details of a corporate project. In another exchange last month, a user asked Claude how to “become become Nine-tailed fox?”, before clarifying they wanted to literally transform from a human into the mythical creature.

Claude initially responded by generating an image claiming the user had been given “fully functional fox powers!”.

Other conversations included users seeking help with resumes that contained names, contact information and employment histories. Some chats also appeared to contain proprietary workplace material, including healthcare research and transcripts of private discussions.

The incident mirrors similar privacy issues involving other AI chatbots. OpenAI last year changed how shared ChatGPT conversations were handled after publicly shared chat logs became searchable online. Grok, the chatbot on Elon Musk’s social media platform X, also had hundreds of thousands of shared conversations indexed by search engines last year.

Google said it does not decide which webpages become publicly searchable.

“We give site owners clear controls to decide whether pages can be crawled or indexed, and we always respect those directives.”

Website owners can prevent search engines from indexing pages using standard web controls or by requesting their removal from search results. Anthropic appears to have used those mechanisms to stop Claude conversation links from appearing in search listings.

Microsoft’s Bing, Brave Search and DuckDuckGo also indexed some of the shared Claude links before they were removed. The companies did not immediately comment.

Stay ahead of the stories shaping our world. Subscribe to Impact Newswire for timely, curated insights on global tech, business, and innovation all in one place.

Dive deeper into the future with the Cause Effect 4.0 Podcast, where we explore the ideas, trends, and technologies driving the global AI conversation.

Got a story to share? Pitch it to us at info@impactnews-wire.com and reach the right audience worldwide