Anthropic Code Leak Sparks AI Shock: Second Time in a Year
Anthropic Code Leak Sparks AI Shock: Second Time in a Year
Security researcher Chaofan Shou uncovers Anthropic's Claude Code leak again via leftover map files, exposing unreleased features and triggering renewed scrutiny.
A new leak of Anthropic's Claude Code tool has rattled the AI industry, with security researcher Chaofan Shou uncovering the issue on March 31. The exposed code surfaced through leftover source map files, laying bare chunks of the development work behind Anthropic's latest AI assistant.
The incident marks the tool's second public leak in about a year, renewing questions about how sensitive AI development assets are protected. The leaked material reportedly includes hints of unreleased features, such as an always-on AI mode and a pet system, underscoring how even seemingly minor scraps can reveal core ideas.
Market chatter and major headlines have connected the episode to broader ambitions around a possible IPO, with some reports claiming the company could be poised for a multi-hundred-billion-dollar public offering. The leak arrives as investors weigh the future of AI startups and the risks of exposing internal engineering work.
Experts warn that source-code leaks can complicate competition, undermine security, and prompt faster regulatory scrutiny for AI vendors. For Anthropic, the latest exposure could intensify calls for stronger safeguards around code, maps, and build environments.
As the AI race accelerates, the industry awaits how Anthropic and others respond—balancing openness with tighter security while continuing to innovate.
