Anthropic, the developer of the Claude AI, recently experienced a significant security incident, accidentally exposing 512,000 lines of its proprietary Claude AI code. The incident occurred due to an improperly included source map, which led to the unintentional release of a substantial portion of its codebase on GitHub. This major leak inadvertently revealed previously unreleased functions and models, offering an unexpected glimpse into the advanced AI system’s inner workings.
Anthropic Accidentally Leaks 512,000 Lines of Claude AI Code on GitHub
By Cedric Ravencroft
A Leeds-based gaming journalist with nine years of experience in the industry. Started covering local gaming tournaments before expanding into national gaming news coverage. Specializes in PC gaming developments and indie game discoveries across the UK. His analytical approach to gaming trends and developer spotlights has earned him recognition among both gamers and industry insiders throughout England