Anthropic emergency control Claude AI agent underlying code leakage incident

After an unexpected leak of the underlying instruction code used to drive the popular AI agent application Claude Code, Anthropic is urgently working to control the impact of the situation. Claude Code has given the company a competitive advantage among developers and enterprise users. As of Wednesday morning, representatives from Anthropic have successfully filed copyright takedown requests, forcing the removal of over 8000 original copies and adaptations of Claude Code instructions shared by developers on the programming platform GitHub. The company later reduced the takedown requests to only 96 copies and adaptations, stating that the initial request covered more GitHub accounts than expected. A spokesperson for Anthropic stated that this "partial internal source code" leak did not involve any user information or data, and did not reveal the highly valuable core algorithm parameters of the company's expensive development of high-performance AI models.