Claude has leaked a slew of user content on Google.
As spotted by Reddit and reported by Futurism, Anthropic’s AI tool has exposed users’ chats and creations in Google search results. It would seem that this is the result of Claude allowing people to share their conversations with others by generating a link that offers public access. However, these links also can be indexed by a search engine like Google, thus revealing their contents to the entire internet.
While Futurism notes that some of the leaked content is harmless, like personal websites and vibe-coded apps, but other material is much more sensitive in nature. This includes a detailed medical reports, names and phone numbers of children, internal company documents and more.
Notably, something similar happened in September when it was discovered that hundreds of Claude chats had been indexed by search engines. Over the past couple of years, we’ve also seen examples of content from other AI tools, like ChatGPT and Grok, getting indexed by Google.
Of course, this wouldn’t be a problem for anyone who (wisely) hasn’t been using Claude or other AI slop tools, but if you happen to be one of them, you might want to double-check that you didn’t get doxxed or anything.
Image credit: Anthropic
