OpenCode, the open source terminal coding agent, disclosed a security breach tied to a large scale distillation attack. In a post on X on October 17, 2026, the project said the incident happened the previous night and impacted 1.4% of OpenCode Console users. Everyone affected was notified by email that morning, and OpenCode said it is working with Anthropic, OpenAI and Google to remediate. The incident is ongoing, so the company has not shared further technical details.
According to notifications received by affected users, unauthorized access hit a legacy Console system that OpenCode was already preparing to shut down. Third party provider keys stored in Console workspaces, including those for OpenAI, Anthropic and Google, may have been exposed. OpenCode said its investigation has found no evidence that other data leaked, and that users who did not receive an email were not impacted. The breach remains unattributed: no post mortem, indicator list or attacker details have been published, and OpenCode has not explained whether the exposed credentials were used to run the distillation campaign it described.
For founders running AI coding agents in production, the practical response is key rotation. Revoke the old credentials in each provider console, issue replacements, update every application that depends on them, and set monthly spend limits or budget alerts so a stolen key cannot quietly burn through your budget. Then check provider usage dashboards and billing for unexpected volume. Teams that only run the OpenCode CLI locally and never linked external providers to Console are outside the stated scope. Distillation, the unauthorized collection of a stronger model's outputs to train a weaker one, is a growing risk for anyone shipping on top of frontier APIs.