Security firm Malwarebytes recently pointed out that shared conversations from Anthropic’s Claude may be indexed by Google Search, potentially exposing sensitive information contained in users’ chats.
According to the report reviewed by IT Home, the security incident was first brought to light by some Reddit users. Through specific search methods, they were able to find a large number of shared Claude conversation records in Google search results.
The issue reportedly stemmed from Claude’s “Share chats” feature. According to Anthropic’s official explanation, the feature generates a publicly accessible link for a conversation selected by the user for sharing and sets the conversation’s status to Public. Malwarebytes found that some publicly shared chat records were crawled and indexed by search engines, eventually appearing in search results.
However, Malwarebytes said this is not an issue unique to Claude. The company had previously also found shared Grok chat records appearing in Google search results, while Meta AI had encountered similar issues in the past.
Malwarebytes noted that completely preventing public web links from being indexed by search engines is not easy. Therefore, to prevent sensitive information from being leaked through shared AI conversations, the simplest and most effective approach is not to publicly share AI chat content with others, while also avoiding entering personal identification information, account passwords, corporate secrets, and other confidential content into AI conversations.