Anthropic inadvertently exposed the source code for its Claude Code AI assistant. A developer accidentally included a source map file in a public software package. The leak exposed over 500,000 lines of proprietary code.
Exposed files detailed the tool's architecture, unreleased features, and internal operations. These details provide a potential roadmap for competitors. An Anthropic spokesperson confirmed the incident resulted from human error rather than a malicious breach.
No sensitive customer data or credentials were compromised. Anthropic removed the package and is implementing new prevention measures. This incident marks the third reported security error for the AI safety-focused startup.