Attackers broke into an older system behind OpenCode Console, the platform where some users of the open source AI coding agent store API keys. About 1.4% of Console users were affected, and developers reported unauthorized charges from hundreds of dollars to more than $1,000. Co-founder Dax Raad said the open source app and command-line tool were not affected.
The breach happened the night OpenCode made GLM-5.3-Flash free for users, and the company disclosed it on Saturday. Anthropic and Google blocked the exposed keys, while OpenAI worked with OpenCode to restore access and refund API customers. OpenCode called the incident ongoing, urged users to revoke and replace their keys, and has not said how attackers got in. It described the attack as a "large-scale distillation attack", a link skeptics say is unproven.
Raad said attackers had probed nonstop for 48 hours, and the AI models his team used for analysis refused to work once the malicious code tripped their safety filters. It follows a similar incident at Hugging Face in July and a recent CERT-In warning about unauthorized AI agent access.