Anthropic accidentally deleted thousands of code repositories on GitHub while attempting to retrieve a copy of the source code for its most popular product from the internet.
On Tuesday, software engineers discovered that Anthropic had accidentally included access to the source code of its category-leading Claude Code command-line application in a recent release. AI enthusiasts scoured the leaked code for clues about how Anthropic uses the application’s underlying LLM and shared it on GitHub.
Anthropic issued a takedown notice under U.S. digital copyright law, asking GitHub to remove the repository containing the offending code. GitHub records show that the notification was sent to approximately 8,100 repositories. It also includes a legitimate fork of the Claude code repository published by Anthropic, according to social media users who were upset that their code was blocked.
Boris Cherny, head of Claude Code at Anthropic, said the move was coincidental and retracted most of the takedown notices, limiting it to one repository and 96 forks containing the mistakenly published source code.
“The repositories named in the notification were part of a fork network connected to our own public cloud code repositories, so the removal spanned more repositories than intended,” an Anthropic spokesperson told TechCrunch. “We have retracted the notification for all but one specified repository, and GitHub has restored access to the affected fork.”
The cleanup failure is another blow to the company, which is reportedly planning an IPO, as it is a task that typically requires careful attention to execution and compliance. As a public company, would you leak your source code? You better believe a shareholder lawsuit is about to happen.
