Users of Anthropic's Claude AI chatbot discovered over the weekend that conversations and Artifacts they had made shareable via a link were being indexed by Google Search, making them publicly accessible. The issue was first reported on Reddit, where a user demonstrated that a Google search for "site:claude.ai/share" surfaced numerous conversations. Soon after, another user found that Artifacts—interactive applications, dashboards, and documents created with Claude—were also appearing in search results. VentureBeat independently verified that some Artifacts were searchable and accessible without authentication. The exposed content included personal discussions, résumés, business materials, and even AI-generated code. Anthropic requires users to explicitly choose to share content and warns that it will be accessible to anyone with the link. However, many users assumed the content would remain unlisted, like a YouTube video, rather than being indexed by search engines. By Sunday morning, many of the original search results appeared to have been removed or became harder to find, suggesting action by Google or Anthropic. The incident echoes similar past issues with OpenAI's ChatGPT and Google's Bard. For enterprises, it highlights the need to clarify what "share" means and to audit shared AI content to prevent unintended exposure.