OpenCode, the popular open-source coding agent used by developers to program with AI models in the terminal and editors, disclosed on October 17, 2026 a security breach related to a large-scale distillation attack.
According to the official statement posted on X by the @opencode account, the incident impacted 1.4% of OpenCode Console users — specifically those who had linked external AI providers such as OpenAI, Anthropic and Google.
What OpenCode reported
The post states that the company notified all affected users by email the following morning. Those who did not receive the email were not impacted. The statement makes clear that the incident was still ongoing at the time of disclosure and that additional details could not be shared at that moment.
OpenCode said it was working jointly with Anthropic, OpenAI and Google to remediate the situation.
What a distillation attack is
Distillation, in the AI context, consists of collecting large volumes of outputs from a more capable model to train or improve another model. Companies such as Anthropic and OpenAI have already described coordinated distillation campaigns that attempt to extract capabilities from frontier models using unauthorized API keys.
In the OpenCode case, the characterization as a “distillation attack” comes from the company itself. User reports mentioned in secondary coverage point to unauthorized charges on provider accounts, but the official statement does not confirm how many keys were effectively accessed or whether there was widespread misuse.
Why this matters
AI coding tools with agents frequently ask the user to connect their own API keys from providers. When those keys are stored in a hosted console or workspace, unauthorized access can generate high costs and credential exposure.
OpenCode is widely used precisely because it is open-source and allows the use of different models. The incident highlights the risk of storing third-party keys in intermediate services, even when the main tool runs locally.
As of the publication of this article, OpenCode had not released a detailed technical post-mortem or publicly confirmed the exact extent of data exposure. Users who received the email were advised to revoke and replace the affected keys in the provider dashboards and to review recent usage.
Sources
Transparency: This content was created, edited or reviewed with the assistance of artificial intelligence. Information was cross-checked with public posts on X and sources available on the internet. Consult the original sources to verify the full context.
By GeekikiBot
