OpenCode Console Breach May Have Exposed OpenAI, Anthropic, and Google API Keys

OpenCode, a tool for working with AI models, says an intruder got into one of its old systems on the night of October 16. The intruder may have reached the AI account keys that some customers saved in its hosted service, called Console. Some users say they have already been charged for usage they never ran.

What happened

Console is an optional service. A team saves its own OpenAI, Anthropic or Google key there, and Console uses that key to call the AI provider. Team members can use the models without ever seeing the key. The breach touched a legacy system, and OpenCode says the keys "may have been exposed." It has not said that any key was copied or used.

Who is affected

OpenCode says 1.4% of Console users are affected, all of them people with an outside provider linked. It emailed them on the morning of October 17 with the subject line "Action Required: Potential provider key breach." Anyone who did not get that email was not affected, the company said on X.

People who use only the OpenCode command-line or desktop apps can relax. Keys added there stay in a database on the user's own machine. Co-founder Dax Raad wrote that the breach involves the Console, not OpenCode itself, and that those keys "never come to us."

Why a saved key is risky

An AI provider key works like a house key that needs no ID. Whoever holds a copy can run billable requests on the owner's account until the provider cancels it. Deleting the key from Console does not cancel it. A spending budget set in Console cannot stop misuse either, because Console only sees requests that pass through its own system.

Users have reported costs. Colin Parsons wrote on X that his OpenAI key was used for $281 and his Anthropic key for $168. Preston Zeller put his losses at $1,700. A Reddit user claimed $400 of unauthorized Anthropic usage, and another claimed about €200 of Google usage. These are the users' own accounts, not confirmed findings.

What is not known

OpenCode called the incident ongoing and said it is working with Anthropic, OpenAI and Google. It also described the breach as "related to a large-scale distillation attack." Distillation is when a campaign pulls huge numbers of answers from an AI model to train a rival one. Anthropic said in September that such groups use fake accounts built on stolen cards, logins and harvested keys.

The company has not said who is behind the attack, how the two connect, what the legacy system was or how it was reached. It has also not said how many keys it held or how many users 1.4% represents.

What to do

If the email arrived, the steps in the alert box above cover what to do. The key point is that the key must be revoked with the AI provider itself, not just removed from Console.

What happens next

The attacks have not stopped. Raad wrote on October 18 that attackers had been "non stop trying to attack our systems for the past 48 hours," and that "a side effect is everything is getting hardened." Affected users should watch their provider billing pages for charges they do not recognize.