Claude shared conversations were indexed by Google and exposed: Sensitive information in AI chats is showing up in search results. Security company Malwarebytes recently pointed out that the shared conversations of Anthropic Claude can be indexed by Google search engines, causing sensitive information in user chats to be accidentally exposed.

Claude’s shared conversation was indexed by Google and exposed: Sensitive information in AI chats is showing up in search results. Security company Malwarebytes recently pointed out that Anthropic Claude’s shared conversations can be indexed by Google’s search engine, causing sensitive information in user chat content to be accidentally exposed. This issue was first discovered by some Reddit users who found a large number of Claude shared conversation records in Google search results through specific searches, including sensitive content such as personally identifiable information and corporate secrets. After investigation by Malwarebytes, it was discovered that the root cause of the problem lies in the "shared chat" function of Claude, which generates publicly accessible links and sets the status to Public for conversations that users choose to share. These public links are then crawled and indexed by search engines, and ultimately exposed in search results. Malwarebytes stresses that this is not an issue unique to Claude. The company has also previously discovered that Grok's shared chat history appeared in Google search results, and Meta AI has experienced similar incidents in the past. This means that the conversation sharing functions of current mainstream AI platforms are at risk of being indexed by search engines, and users may make private conversations public without their knowledge. Malwarebytes pointed out that it is not easy to completely prevent public web links from being indexed by search engines, and there are inherent limitations in technical protection. Privacy Blind Spots of AI Conversation Sharing Function The shared conversation function is intended to allow users to conveniently display the results of AI interactions, such as sharing interesting answers or showing work processes. However, there is a huge gap between the "public" status setting and users' psychological expectations. When most users click the "Share" button, they don't realize that the generated link will be permanently included by search engines, and they don't expect that the content will appear in anyone's search results. The peculiarity of AI conversations is that users tend to enter real work content and personal information into them, and once indexed, the information leaked is far more sensitive than ordinary social media posts. The protection suggestions given by Malwarebytes are direct and effective: do not share AI chat content publicly, and avoid entering confidential content such as personally identifiable information, account passwords, and corporate secrets in AI conversations. However, this suggestion essentially requires users to change their usage habits, rather than solving the problem at the platform level. More reasonable solutions should include: setting shared links to be excluded from inclusion by default, providing password protection options, and clearly informing users of privacy risks before sharing. As the amount of information carried by AI conversations increases, platforms have the responsibility