At a Glance · AI-generated

Anthropic accidentally leaked the source code for Claude Code, its AI coding tool, on March 31 due to an npm packaging error. The leak revealed internal features including a "Tamagotchi-style pet" and system data collection capabilities. In response, Anthropic removed thousands of GitHub repositories in an attempted cleanup that the company later acknowledged was itself accidental. Despite the takedown efforts, the code was cloned over 8,000 times and subsequently exploited to distribute malware by April 3. This incident marked the company's second major unintentional disclosure in days, following a leak of its unreleased "Mythos" model.