Claude AI chat logs exposed in search results, now removed

Hundreds of user conversations with Anthropic's Claude AI chatbot were discoverable through Google and other search engines until this past weekend, raising fresh questions about how AI companies handle shared content.

The issue affected chat logs that users had explicitly chosen to share via Claude's built-in sharing feature. These links were indexed by search engines, making them accessible to anyone using site-specific search terms. The exposed conversations included personal information, work details, resumes with contact information, and what appeared to be proprietary research.

How the exposure happened

Reddit users first identified the publicly available chats, which spanned more than 200 conversations across at least 25 pages of search results. Some conversations were as recent as a few weeks before discovery.

An Anthropic spokeswoman told the BBC that users maintain control over sharing and that conversation links are "not guessable or discoverable unless people choose to share them themselves." She noted that when users share a conversation, they make that content publicly accessible, and like other public web content, it may be archived by third-party services.

The share feature in Claude informs users that "anyone with the link" can view the contents, but does not explicitly state that the link may appear in search engine results.

What was exposed

The indexed conversations covered a wide range of topics. In one April conversation, a user asked Claude to draft an unpublished blog post about cloud security involving corporate project details. Other users sought help with resumes, including names, contact information, and work history. Some conversations appeared to involve proprietary healthcare research with transcripts of private discussions.

Anthropic removed the search availability of these chat logs over the weekend, likely using standard tools that allow website owners to block links from search results. A Google spokesman clarified that the company does not control what pages are made public on the web, noting that websites themselves must use available controls to decide whether pages can be crawled or indexed.

Why it matters

This incident highlights a critical gap in how AI platforms communicate data exposure risks to users. While Anthropic's sharing feature technically functioned as designed, many users likely did not anticipate their conversations would be indexed by search engines and discoverable by the public. For enterprises evaluating AI tools, this underscores the need for clear policies around data sharing and explicit employee training on privacy implications when using generative AI platforms.

A recurring pattern

This is not an isolated incident in the AI industry. OpenAI experienced a nearly identical issue with ChatGPT chat logs being made publicly accessible last year, ultimately changing how easily such logs could be shared. Grok, the AI chatbot on Elon Musk's X platform, also saw hundreds of thousands of chat logs made publicly available through online search last year.

The details were first reported by the BBC.