Thousands of Claude Chat Logs Exposed Online Leaking Highly Personal Data
Anthropic
A large collection of Claude AI chat logs containing sensitive personal information has been found publicly accessible on the internet. The exposed data includes details about users' medical conditions, sexual fantasies, and proprietary company information.
Futurism reports that a massive dataset of conversations with Anthropic's Claude AI assistant is publicly accessible online, exposing extremely private user information. The leaked chats contain details about users' medical conditions, sexual fantasies, company source code, and other sensitive data. This breach occurred because users or third parties made their chat logs publicly available on shareable links, which were then indexed by search engines or scraped by web crawlers. The exposed information includes not only personal confessions but also proprietary business information, such as an entire PHP backend with API keys. The incident raises serious privacy concerns as users may have assumed their conversations were private.
Source: Anthropic (GNews) —
original
