Hundreds of private, personal, and sometimes slightly disturbing prompts and chats made with Claude.AI surfaced online through nothing more than a search.
Maybe it was a novelist figuring out a plot, otherwise if you’re in Kansas and you have an attorney, perhaps you should be the one asking questions.
This isn’t like the way Claude Cowork recently escaped its sandbox and went through users’ files, and it also isn’t still a problem. At time of writing, all signs of this privacy breach have been removed.
But according to Wired, for at least two days, private chats made with Anthropic’s Claude were readily available online. It was not possible to zero in on a specific person, but Redditors found that a simple search for “site:claude.ai/share” surfaced hundreds of such chats.