Anthropic took down thousands of GitHub repos trying to yank its leaked sourcecode — a move the company says was an accident

Anthropic accidentally caused the takedown of around 8,100 GitHub repositories while trying to remove leaked source code of its Claude Code application. The error was attributed to a misapplied takedown notice, affecting legitimate forks of its own repository. The company has since retracted most notices, restoring access, though the incident raises concerns as Anthropic prepares for an IPO.

Key Points

• Anthropic's source code for Claude Code was leaked online, discovered by a software engineer.
• Thousands of AI enthusiasts examined the leaked code, sharing it on GitHub.
• Anthropic issued a takedown notice under U.S. copyright law, targeting approximately 8,100 repositories.
• The notice affected legitimate forks of their own public repository, leading to user backlash.
• Anthropic acknowledged the error and retracted most takedown requests, restricting it to one repository and 96 forks.

Relevance

• This incident highlights challenges tech companies face in managing intellectual property, especially in the rapid expansion of AI.
• Similar past incidents include Oracle's copyright issues over Java or GitHub’s policies on code ownership.
• As of 2025, trends indicate increased scrutiny on AI transparency and compliance, making this incident pertinent to ongoing discussions on corporate responsibility.

The accidental takedown reflects potential mismanagement at Anthropic as it prepares for an IPO, emphasizing the need for robust compliance in handling source code and public repositories.