Hackers Breach AI Coding Tool OpenCode, Causing Unauthorized Charges on Users' Accounts
OpenCode, an open-source AI coding tool used by developers as an alternative to platforms like Claude Code and Cursor, reported a security breach affecting 1.4% of its users who store API keys in the OpenCode Console platform. The attack involved a large-scale distillation process that led to unauthorized charges on some users' accounts with major AI providers such as OpenAI, Anthropic, and Google. Developers began reporting unexpected spending alerts and hundreds — or even thousands — of dollars in unapproved usage fees overnight. OpenCode confirmed it had notified affected users via email but did not disclose the exact method used for the breach. The exposed API keys were limited to Console workspaces; OpenCode's open-source app and command-line tool remained unaffected, according to co-founder Dax Raad. While some skeptics questioned whether this was a true distillation attack or simply an overblown breach report, major AI companies like Anthropic and Google took action by blocking compromised keys. The incident highlights the growing challenge for smaller firms without vetted access to advanced AI models in defending against sophisticated cyberattacks. OpenCode's team had to rely on defensive agents running continuously during a 48-hour attack window but faced limitations when safety filters blocked their analysis of malicious code.
Topics: AI API keys data breach distillation attack open-source