OpenCode, the open-source AI coding agent, disclosed a security breach on October 17, 2026, saying attackers got unauthorized access to a legacy system and may have exposed the OpenAI, Anthropic and Google API keys that some users had saved in their workspaces. The company calls it a large-scale distillation attack and says about 1.4% of OpenCode Console users were affected. Why it matters: anyone who linked a provider account to OpenCode Console should rotate those keys today, because a stolen AI API key can run up real bills fast.
October 17, 2026
What happened at OpenCode?
OpenCode said it identified unauthorized access to a legacy system that was already in the process of being decommissioned. In an email to affected customers and a public post on X, the company said third-party provider keys saved in user workspaces may have been exposed, and that everyone affected was notified individually by email. If you did not get that email, OpenCode says you were not among those impacted. The affected group is a subset of Console users who had external provider accounts linked.
The company has shared few technical details so far. It says the incident is ongoing, that it cannot share more at this point, and that it is working with Anthropic, OpenAI and Google on remediation. Co-founder Dax Raad also posted that his personal accounts were being attacked, calling it a “whole new world” of AI-powered hacking.
What is a distillation attack, and why do the keys matter?
Distillation normally means training a smaller model on a bigger model’s outputs. In attack form, it means systematically querying a frontier model at scale — ideally on someone else’s paid account — to harvest responses that can be used to train a rival model. That is why the target here was not passwords or card numbers. It was live keys to some of the most expensive models on the market.
For affected users, the practical risk is immediate and financial: a working key can be hammered with queries until the owner notices. Some affected users have reported balances draining quickly, with bills climbing into the hundreds of dollars. OpenCode has not said how unauthorized charges will be handled.
What should OpenCode users do now?
OpenCode’s advice is straightforward. Revoke and replace every provider key that was linked to OpenCode, through each provider’s own dashboard. Update any application, script or tool that used the old keys. Then review recent usage on your OpenAI, Anthropic and Google accounts for anything you do not recognize.
Even if you were not notified, this is a useful prompt to check where your keys live. OpenCode is popular precisely because it is provider-agnostic — it connects to more than 75 model providers — but “bring your own key” tools concentrate several high-value credentials in one place. One breach at the intermediary can expose all of them at once.
Why it matters
This is a new shape of breach for the AI era. The valuable asset was not stored personal data; it was metered access to frontier models, billed by the query. As labs tighten direct access to their best models, intermediaries that hold other people’s keys become more attractive targets. Our take, in the spirit of everything AI, tested: treat an AI API key like a payment card. Store as few as possible in third-party dashboards, set spending limits where providers offer them, and rotate keys the moment a vendor tells you to — not next week.
Frequently asked questions
Was I affected by the OpenCode breach?
OpenCode says it emailed every affected user individually, and that users who received no email were not impacted. The company puts the affected group at roughly 1.4% of OpenCode Console users who had external providers linked.
Which API keys may have been exposed?
Keys for third-party AI providers saved in OpenCode workspaces, with OpenAI and Anthropic named as examples in the notice and Google also reported among the affected providers.
What should I do if I linked keys to OpenCode?
Revoke and replace the keys in each provider’s dashboard, update anything that used the old keys, and check recent usage for unexpected activity.
Did OpenCode say whether the keys were actually used?
Not in the details published so far. The company says the incident is ongoing and that it will share further updates as its investigation continues.
Sources: OpenCode (company post and customer notice, via X), OfficeChai, Crypto Briefing, RuntimeWire