Anthropic's leaked AI coding tool has been cloned over 8,000 times on GitHub despite mass takedowns

Anthropic's leaked Claude Code source prompted copyright takedowns of over 8,000 GitHub copies and adaptations.

A programmer used AI tools to rewrite the leaked code in other languages and kept versions available despite removals.

The leak exposed a "harness" and a "dreaming" function that show how Anthropic controls models and could be replicated by competitors.

The incident follows a separate Mythos leak and occurs while Anthropic plans an IPO at a $380 billion valuation.