What is the story about?
Anthropic has found itself at the centre of another controversy after inadvertently causing thousands of code repositories on GitHub to be taken offline while attempting to remove leaked source code from its popular Claude Code tool.
The incident unfolded earlier this week when a software engineer discovered that a recent release had unintentionally exposed parts of the internal source code for Claude Code, a command-line application built around Anthropic’s large language model technology. The leak quickly spread across developer communities, with users analysing and reposting the code on GitHub to better understand how the system operates.
Anthropic responded by issuing a takedown request under US copyright law, seeking to remove repositories hosting the leaked material. However, the move had unintended consequences, sweeping up far more content than originally intended.
According to GitHub records, the takedown notice impacted around 8,100 repositories. This included not only copies of the leaked code but also legitimate forks of Anthropic’s own public Claude Code repository, prompting backlash from developers whose projects were suddenly made inaccessible.
The broad scope of the removal appears to have been linked to how GitHub handles “fork networks”, where related repositories are grouped together. As a result, the takedown extended beyond the targeted content, affecting a much wider set of projects.
Anthropic later acknowledged the mistake and moved quickly to reverse most of the removals. Boris Cherny, head of Claude Code, said the action was unintentional, and the company worked to restore access to affected repositories, reports TechCrunch.
The chain of events began when the source code was accidentally included in a software release, making it accessible to users. The code related to Claude Code, one of Anthropic’s key developer tools, which has gained popularity for its ability to assist with coding tasks.
Once discovered, the leak attracted significant attention from AI enthusiasts and developers, who examined the material for insights into how Anthropic’s models are deployed in practical applications. The rapid spread of the code online made containment difficult, prompting the company’s aggressive takedown response.
While Anthropic has since corrected the overreach, the episode adds to a string of recent missteps for the company. It also comes at a sensitive time, as reports suggest Anthropic is considering a public listing, where operational precision and compliance are closely scrutinised.
The incident highlights the challenges tech firms face in managing intellectual property in the age of open collaboration, where code can be quickly shared and replicated across platforms.
The incident unfolded earlier this week when a software engineer discovered that a recent release had unintentionally exposed parts of the internal source code for Claude Code, a command-line application built around Anthropic’s large language model technology. The leak quickly spread across developer communities, with users analysing and reposting the code on GitHub to better understand how the system operates.
Anthropic responded by issuing a takedown request under US copyright law, seeking to remove repositories hosting the leaked material. However, the move had unintended consequences, sweeping up far more content than originally intended.
Anthropic took down thousands of GitHub repos
According to GitHub records, the takedown notice impacted around 8,100 repositories. This included not only copies of the leaked code but also legitimate forks of Anthropic’s own public Claude Code repository, prompting backlash from developers whose projects were suddenly made inaccessible.
The broad scope of the removal appears to have been linked to how GitHub handles “fork networks”, where related repositories are grouped together. As a result, the takedown extended beyond the targeted content, affecting a much wider set of projects.
Anthropic later acknowledged the mistake and moved quickly to reverse most of the removals. Boris Cherny, head of Claude Code, said the action was unintentional, and the company worked to restore access to affected repositories, reports TechCrunch.
Claude Code leak: What happened
The chain of events began when the source code was accidentally included in a software release, making it accessible to users. The code related to Claude Code, one of Anthropic’s key developer tools, which has gained popularity for its ability to assist with coding tasks.
Once discovered, the leak attracted significant attention from AI enthusiasts and developers, who examined the material for insights into how Anthropic’s models are deployed in practical applications. The rapid spread of the code online made containment difficult, prompting the company’s aggressive takedown response.
While Anthropic has since corrected the overreach, the episode adds to a string of recent missteps for the company. It also comes at a sensitive time, as reports suggest Anthropic is considering a public listing, where operational precision and compliance are closely scrutinised.
The incident highlights the challenges tech firms face in managing intellectual property in the age of open collaboration, where code can be quickly shared and replicated across platforms.
