Hackers stole OpenCode users' AI keys — some bills topped $1,000

Attackers broke into an older OpenCode system overnight, the company said Saturday, reaching API keys that some users had stored in OpenCode Console, its hosted platform. The breach affected 1.4% of Console users, and the company emailed all of them. The open-source app and command-line tool were not affected — co-founder Dax Raad noted that the keys never come to OpenCode itself.

Developers woke to spending alerts firing across their OpenAI, Anthropic and Google accounts. Some reported unauthorized charges in the hundreds of dollars, and a few described totals above $1,000. OpenAI said it worked with OpenCode to restore access and refund credits used in the attack; Anthropic and Google blocked the keys they believed were exposed, and OpenCode says refunds for unauthorized charges are underway.

OpenCode described the incident as a large-scale distillation attack — stolen keys used to generate model outputs at volume — though some skeptics called that framing an overstretch. The defense hit its own wall: the models the team used to analyze the attack began refusing once the malicious code tripped safety filters, until OpenAI granted the team access to its Daybreak Blue program for vetted defensive security work.