Anthropic’s AI code leak ignites frenzy among Chinese developers

Anthropic, a US AI company that restricts access to its services in China due to national security concerns, inadvertently leaked the source code for its AI coding tool, Claude Code. The leak occurred when an employee accidentally included the modified code in a software package hosted online. Shou Chaofan, a software engineer, discovered and decrypted the 512,000-line code, posting it on X.com, where it gained widespread attention. Despite Anthropic's restrictions, Chinese developers are highly interested in Claude Code's advanced capabilities, which automate programming tasks and boost efficiency. The incident has sparked a frenzy among Chinese developers eager to analyze and utilize the leaked code.

Anthropic accidentally leaked part of the internal source code for its AI coding assistant, Claude Code, due to human error during a software update. The leaked archive contained nearly 2,000 files and 500,000 lines of code, which were quickly copied to GitHub, prompting copyright takedown requests from Anthropic. The exposed code revealed blueprints for features like a Tamagotchi-esque coding assistant and an always-on AI agent. While Anthropic claims no sensitive customer data was exposed, this is the second data leak for the company in recent weeks, raising concerns about internal security vulnerabilities. The leak could potentially benefit competitors like OpenAI and Google.