Hackers took API keys from 1.4% of OpenCode Console users

OpenCode, an open source AI coding agent that has become a popular alternative to Claude Code and Cursor, said attackers broke into an older system overnight. About 1.4% of users of its online OpenCode Console were affected. Some store the API keys they use for AI services there.

Developers posted spend alerts and unauthorized charges. Some reported hundreds of dollars on individual accounts and a few reported totals above $1,000. Anthropic and Google blocked the keys they believed were exposed, OpenAI said it worked with OpenCode to restore access and refund its API customers, and OpenCode says refunds are underway. The open source app and the command-line tool were not affected, and the company has told affected users to revoke and replace their keys with each provider.

The team said attackers kept trying to get in for 48 hours. AI models it asked for help analyzing the malicious code started refusing to work, because the code tripped their safety filters. OpenCode later got access to a program that gives vetted users models with fewer such refusals for defensive security work.

OpenCode called it a distillation attack. Some outside observers say it was a plain breach and that the distillation claim is unproven.

Our take

The scary part is not the stolen keys, it is the defenders being told no by their own tools. Guardrails that cannot tell an incident responder from an attacker slow down the people under attack.