OpenCode has notified users that stored API keys for external AI providers were exposed in a security breach. The project maintainer said the incident affected 1.4% of OpenCode Console users, all of whom had linked external providers.
OpenCode is an open-source AI coding agent. Developers use natural-language instructions to understand code, write features and fix bugs through a terminal, desktop app or code editor. Its Console lets users connect models from providers including Anthropic, OpenAI and Google.
OpenCode announced the breach on Saturday and said it was related to a large distillation attack. The maintainer said affected users were notified by email that morning.
Distillation trains smaller AI models on responses from larger ones. It can be used legitimately to build cheaper, faster models. Large-scale training through unauthorized queries makes it an attack.
Read nextNvidia releases open models for enterprise agents and local PCsAttackers targeted OpenCode with AI agents
OpenCode co-founder Dax Raad said attackers had targeted the company's systems with AI agents for at least 48 hours. He said OpenCode's investigation was hindered when the AI models it used began refusing requests because of built-in security safeguards. Raad said the situation stabilized after OpenCode received approval on Saturday to use OpenAI's Daybreak AI cybersecurity model for defensive work.
According to the notification sent to affected users, their keys could have been exposed as early as October 17, 2026, at 00:30 UTC. The email also said no data appears to have been exposed.
Users reported unauthorized API charges
Multiple users reported unauthorized API charges, ranging from several hundred dollars to more than $1,700 overnight. Some said Google suspended their projects after detecting “abusive activity consistent with hijacked resources.” OpenCode's message to affected users raised the possibility of reimbursement.
BleepingComputer contacted OpenCode for more details but had not received a response by publication. OpenCode said it is working with Google, OpenAI and Anthropic to address the incident and cannot share further details yet.
