- They thought they were downloading Claude Code source. They got a nasty dose of malware instead The Register —
- House Democrat pushes Anthropic on safety protocols, source code leak The Hill —
- Anthropic says its leak-focused DMCA effort unintentionally hit legit GitHub forks Ars Technica —
- Anthropic CCO on Claude code leak on GitHub: It was result of 'process errors' Times Of India —
- How an engineer ensured Claude Code source code leak stays on GitHub despite Anthropic's takedown notice Times Of India —
- Claude’s code: Anthropic leaks source code for AI software engineering tool The Guardian —
- Claude Code bypasses safety rule if given too many commands The Register —
- Anthropic took down thousands of GitHub repos trying to yank its leaked source code — a move the company says was an accident Techcrunch —
- Here's what that Claude Code source leak reveals about Anthropic's plans Ars Technica —
- Anthropic’s Claude Code Leak Revealed Unreleased Features Bloomberg —
Anthropic Admits to Claude Code Leak
The leak involved nearly 2,000 internal files and occurred during a rapid product release cycle, highlighting the security risks in the competitive AI landscape.
While Anthropic initially hit legitimate forks with takedown notices, executives have since retracted the bulk of those claims, calling them accidental.
The incident has revealed unreleased features, including a persistent agent and a virtual assistant named 'Buddy.' This security lapse comes as Anthropic navigates a legal dispute with the Pentagon over supply chain risk designations.