OpenCode Users Face API Key Exposure Risk After Legacy System Breach Tied to Distillation Attack

Organisations using OpenCode Console face potential credential exposure after the developer disclosed a security breach affecting a small subset of its user base. The incident involves third-party API keys stored in workspaces, meaning that access tokens for external services could be compromised. This creates immediate risk for any applications or integrations relying on those specific keys.

Who is affected by this incident?

The breach impacts 1.4% of OpenCode Console users, specifically those who had linked external providers to their workspaces. According to Cyber Security News, this percentage refers only to Console users, not the entire OpenCode customer base. OpenCode stated it emailed every affected individual on the morning of 17 October, clarifying that anyone who did not receive such a notification was not impacted by the breach.

View the original post on X

What data was potentially exposed?

Third-party API keys saved in affected workspaces may have been exposed, including credentials for major AI providers such as OpenAI and Anthropic. The notification indicates that unauthorised access occurred on a legacy system that OpenCode was already preparing to shut down. The company’s wording describes possible exposure rather than confirming that every affected key was definitively stolen or actively used by attackers.

How did the breach occur?

OpenCode linked the incident to a large-scale distillation attack, a method where attackers train one AI model using responses generated by another. The company stated it is working with Anthropic, OpenAI, and Google to address the issue. However, the public statement does not explain how attackers gained initial access, how long they remained inside the system, or provide technical evidence showing exactly how the breach supported the reported distillation activity.

What is the current status?

There is no confirmed financial loss figure in the public disclosure, and OpenCode has not named a specific attacker. The company has not explained whether the exposed credentials were actually used to make unauthorised requests. The immediate response for notified users involves the key replacement process described in the warning email, while the broader investigation continues in coordination with the AI providers.

What affected people can do

Teams should immediately verify if they received the notification email from OpenCode. If so, check dashboards for OpenAI, Anthropic, and other linked providers to revoke any potentially compromised API keys. Generate new credentials and update all applications that depend on them. Monitor recent usage logs for unexpected activity or unusual request patterns. Be vigilant against follow-up phishing attempts that may reference this breach to trick users into revealing new credentials or personal information.

Background: Initial access

Initial access is rarely a single technical failure. It usually combines social engineering with misconfigured identity settings. You must stop viewing it as a perimeter breach and start treating it as an identity and trust validation problem.

Read the full guide: Initial Access Misconceptions That Leave Your Network Exposed

What to do and how to stay safe: OpenCode

• Audit all third-party API keys stored in your development environments and rotate any that have not been recently changed.
• Review access logs for external services linked to your workspaces to identify any unauthorised or anomalous usage patterns.
• Ensure that legacy systems or deprecated infrastructure are fully decommissioned and disconnected to reduce the attack surface.
• Educate staff on recognising phishing emails that reference recent security incidents to harvest new credentials or sensitive data.

Step-by-step guide: Identity Theft Protection: Answers to Critical Security Questions

General security guidance from the Phish Herald newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

Did OpenCode confirm that all API keys were stolen?

No, OpenCode stated that keys "may have been exposed" but did not confirm that every affected key was stolen or used.

How can I know if I am affected?

OpenCode emailed all affected users on 17 October; if you did not receive an email, you were not impacted.

Which AI providers were involved?

The incident potentially exposed keys for OpenAI, Anthropic, and Google, with which OpenCode is working to address the issue.