The AI assistant Claude, crafted by Anthropic, has come under scrutiny after it was discovered that some of its chat records are being indexed by search engines. A considerable volume of user dialogues, along with Artifacts harboring confidential details, have surfaced in Google search outcomes. This predicament is intricately tied to Claude's "Share Chat" functionality. Anthropic has clarified that such content is only indexed when users deliberately share the links in accessible online spaces, emphasizing that the company does not actively submit content to search engines. At present, measures have been implemented to mitigate the issue to some degree. This event highlights the critical need for robust public sharing mechanisms, meticulous data permission management, and effective search engine indexing controls in the security architecture of AI products. Users can review publicly shared chat links within Claude's designated settings. It's worth noting that ChatGPT encountered similar issues with publicly shared conversations being crawled before. Furthermore, among the content exposed in this instance, there was also pornographic material, which clearly violates Anthropic's usage policies.