Anthropic inadvertently led to the removal of thousands of code repositories on GitHub while attempting to retrieve copies of the source code for its most popular product from the internet.

On Tuesday, a software developer found that Anthropic had, apparently by mistake, granted access to the source code for the leading Claude Code command line application in a recent version. AI fans examined the leaked code for insights on how Anthropic utilizes the LLM underpinning the application, disseminating it on GitHub.

Anthropic submitted a takedown request under U.S. digital copyright law, urging GitHub to remove repositories containing the problematic code. GitHub’s records indicated the notice was executed against approximately 8,100 repositories — including legitimate forks of Anthropic’s publicly available Claude Code repository, as expressed by frustrated social media users whose code was affected.

Boris Cherny, Anthropic’s head of Claude Code, stated that the action was unintentional and retracted most of the takedown requests, restricting it to a single repository and 96 forks containing the mistakenly released source code.

“The repository mentioned in the notice was part of a fork network tied to our own public Claude Code repository, so the takedown impacted more repositories than planned,” an Anthropic representative informed TechCrunch. “We have retracted the notice for all but the one repository we specified, and GitHub has reinstated access to the affected forks.”

This mismanaged clean-up reflects poorly on the company as it reportedly prepares for an IPO, a process that usually requires a focus on execution and compliance. Leaking your source code as a publicly traded company? You can bet a shareholder lawsuit is on the horizon.