Private Claude chats exposed in Google and Bing search results

Private conversations with Anthropic's Claude AI chatbot were inadvertently made publicly accessible through Google and Bing search results, exposing sensitive user data. The incident occurred when web crawlers indexed shared links to Claude chats that were meant to be private, allowing anyone to view the discussions by searching for specific phrases or URLs.

The affected users include individuals who used Claude's "share chat" feature, which generates a unique link intended for private sharing. However, these links were not properly blocked from search engine indexing, leading to the exposure of conversations containing personal information, business strategies, and other confidential details. The breach highlights the vulnerability of users who trusted the platform's privacy controls.

This matters because it undermines confidence in AI chatbot privacy protections, especially as tools like Claude are increasingly used for sensitive tasks such as therapy, legal advice, and corporate planning. The exposure could lead to reputational damage, identity theft, or competitive intelligence leaks for affected individuals and organizations.

According to the source, the issue was discovered when security researcher Johann Rehberger found that Claude chat links were appearing in Google and Bing search results. Anthropic confirmed the problem on March 14, 2025, stating that a small percentage of shared links were affected. The company did not disclose the exact number of exposed chats but said it was working to remove the links from search engines.

Background: This is not the first time AI chatbots have faced privacy issues. In 2023, Samsung employees accidentally leaked trade secrets via ChatGPT, and similar incidents have occurred with other chatbots. Anthropic had previously implemented measures to prevent indexing, but the latest incident suggests those safeguards were insufficient.

Anthropic has since updated its robots.txt file to block crawlers from accessing shared chat links and is contacting affected users. The company advises users to avoid sharing sensitive information in chats and to delete shared links after use. The incident may prompt stricter regulations on AI data handling and push companies to adopt more robust privacy defaults.