Anthropic inadvertently triggered the removal of around 8,100 GitHub repositories while attempting to erase traces of its leaked source code. The incident began when a software engineer found that a recent release had mistakenly exposed the source code for Claude Code, the company's widely used command line tool powered by its large language model. Once the code appeared online, developers uploaded copies to GitHub, prompting Anthropic to issue a takedown request under U.S. copyright law. However, due to the interconnected nature of GitHub's fork system, the enforcement action swept up numerous unrelated or legitimate derivative projects, including forks of Anthropic's own open-source repository. The company's head of Claude Code, Boris Cherny, confirmed the overreach was unintentional. An Anthropic spokesperson told TechCrunch the takedown "reached more repositories than intended" because the targeted repo was linked to the public Claude Code fork network. The company quickly retracted most of the notices, narrowing the action to just one repository and 96 forks that contained the sensitive code. GitHub has since restored access to the wrongly affected projects. The mishap comes at a sensitive time for Anthropic, which is reportedly preparing for an IPO, where operational precision and legal compliance are under heightened scrutiny.
When Anthropic says the mass takedown was a mistake, that reveals a deeper issue: the company's apparent lack of control over its own release and enforcement systems. Pulling down 8,100 repositories — including its own open forks — suggests automation or legal processes running unchecked, a red flag for a firm eyeing public markets. For AI firms like Andela or Nigerian startups experimenting with LLM tooling, this is a lesson in how fast intellectual property missteps can spiral — especially when open-source norms collide with proprietary ambitions. Trust erodes not when code leaks, but when the fix causes more damage than the flaw.
