Claude Code Source Leak Raises Security Questions at Anthropic

The Rundown

The Guardian reports that Anthropic briefly exposed internal source code and files related to Claude’s software engineering tooling after a human error, before the materials were taken down. The incident highlights how sensitive AI infrastructure has become and how costly operational mistakes can be in the race to ship agentic developer tools.

The Details

• •The Guardian says nearly 2,000 internal files were briefly exposed due to human error
• •The leak involved code tied to Anthropic’s AI software engineering tooling
• •The files were reportedly discovered and circulated before being removed
• •It reinforces that security and release processes are now a core competitive risk for AI labs