Anthropic Accidentally Wiped Thousands of GitHub Repos While Trying to Contain Claude Code Leak

Mass Repository Takedown Following Code Leak

AI startup Anthropic inadvertently caused the removal of approximately 8,100 repositories on GitHub while scrambling to contain a leak of its Claude Code source code, according to TechCrunch.

Anthropic confirmed the partial source code leak of its developer tool on March 31. A company spokesperson stated that no confidential customer data was exposed and that the leak resulted from a human error during release packaging — not a security breach. The company said it is implementing measures to prevent a recurrence.

The incident poses a competitive risk for Anthropic, as rivals and independent developers gained full access to the architecture of one of its flagship products.

Why This Matters

This incident sits at the intersection of several sensitive issues in the tech industry. The leak of a commercial AI product's source code gives competitors a detailed look at its internal architecture. Meanwhile, the mass removal of GitHub forks at a single company's request raises questions about the balance between copyright enforcement and the rights of the open-source community. Thousands of projects that were merely part of a fork network — with no direct connection to the leak — were caught in the crossfire.

Takedown Backlash and Reversal

Anthropic filed takedown requests with GitHub under U.S. copyright law, targeting repositories containing the leaked code. The sweep affected roughly 8,100 archives, including forks of the official Claude Code repository. The action triggered widespread backlash from affected users whose legitimate projects were collateral damage.

Boris Cherny, who leads the Claude Code team, publicly acknowledged the misstep:

"The repositories listed in the notice were part of a fork network connected to our public Claude Code repository. The removal affected more projects than intended. We retracted the notices for all but one repository, and GitHub has restored access to the affected forks." — Boris Cherny (@bcherny), original post

Anthropic ultimately withdrew the vast majority of its takedown requests, and GitHub reinstated access to the blocked projects.

Community Launches OpenClaude Using Leaked Code

While the controversy unfolded, enthusiasts launched an open-source project called OpenClaude, built on top of Anthropic's leaked code. The fork includes support for Anthropic's API as well as models compatible with the OpenAI interface. Users can connect solutions from Google (via OpenRouter), Mistral, OpenAI, or free local models including those served through Ollama.

OpenClaude retains the original Claude Code toolset: web search, file reading and editing, screenshot analysis, and cross-session memory. Additionally, the developers introduced a router that automatically selects an AI model based on task complexity. However, the deep reasoning mode available in the original Claude Code does not currently work in OpenClaude.

For context, Anthropic turned Claude into a full-fledged AI agent in March, giving the bot the ability to operate a computer autonomously to complete user tasks.