• 2 min read
Google indexed shared Claude chats with sensitive data
Google indexed shared Claude chats containing medical records and confidential documents, after users posted share links on crawlable sites.
Image: iXBT
Google search results are exposing a vast number of Claude chats and artifacts—interactive mini-apps and documents users can create inside Anthropic’s chatbot. Reddit users found the material by searching Google for queries such as site:claude.ai/share.
Some of the indexed conversations reportedly contained medical records, confidential company documents, and the names and phone numbers of children. The discovery was reported on July 29, 2026.
“Anyone with the link can view it.”
How Claude’s sharing links became searchable
The issue is tied to Claude’s “share chat” feature. It lets users generate a URL that anyone with the assigned link can use to view a conversation or project. The interface warns users about that access level, but the feature appears intended for sharing with friends, colleagues, or small groups—not for publishing material to the entire internet.
Recommended reading
Zuckerberg wants superintelligence available to everyone
Anthropic told TechCrunch that a shared link appears in search results only when it has been posted somewhere accessible to search engines, such as a forum or social-media post. The company added that a link sent privately to another person does not appear in search results.
That behavior differs from the practical outcome users might expect from comparable services. Google Docs also supports shareable links, but documents shared that way do not consequently become publicly searchable through Google.
Anthropic has blamed users for exposing the information, arguing that the links had to be published in crawlable locations before Google could index them. The company did not say how many Claude chats or artifacts were affected, nor did it announce a removal date for the indexed results.
Similar exposure at DeepSeek
The incident follows a comparable disclosure involving DeepSeek. On July 22, publicly accessible conversations from customers of the Chinese chatbot appeared online, including work documents and private messages. Those records were also reported to have reached the internet through DeepSeek’s sharing option.
The DeepSeek case had a direct workplace consequence: an executive at a Moscow-based company was fired for disclosing a secret through the chatbot.
AI Editor
Ava covers the rapidly evolving world of artificial intelligence, from foundational models and research labs to the real-world economics of intelligence. With a background in computational linguistics, she cuts through the hype to find out what actually works. She firmly believes that benchmarks are just marketing until reproduced in the wild.
via iXBT
