TLDR:
- OpenCode says the breach affected approximately 1.4% of Console users with external providers linked.
- Exposed credentials may include API keys for OpenAI, Anthropic, and Google.
- The company identified a large-scale distillation attack but has not disclosed full technical details.
- Affected users should revoke keys, replace credentials, update applications, and review recent API usage.
OpenCode disclosed a security breach that exposed third-party API keys belonging to approximately 1.4% of its Console users. The incident involved unauthorized access to a legacy system that the company was preparing to decommission.
Exposed credentials may have included keys for OpenAI, Anthropic, and Google, according to OpenCode’s customer notification. The company said attackers carried out a large-scale distillation attack, although it has not released further technical details.
Affected users were instructed to revoke their keys, replace them, and check for unexpected usage and charges.
The disclosure raises concerns about credential security across AI development tools. Users who connect external providers through third-party platforms may face financial risks when attackers obtain their credentials.
How the OpenCode Security Breach Exposed API Keys
OpenCode announced the incident on X, confirming that unauthorized access affected a subset of Console users with external providers linked. The company said it notified all identified affected users by email on the morning following the breach.
The customer notice identified a legacy system as the entry point. OpenCode had been preparing to decommission that system when the unauthorized access occurred. Third-party provider keys stored in user workspaces may have been exposed.
The company is working with OpenAI, Anthropic, and Google to address the incident. However, OpenCode said the investigation remains ongoing and declined to share additional details.
Security commentator Matt Johansen also reported customer concerns about stolen keys and unexpected API charges. Community discussions described users receiving breach notifications and warnings to revoke their credentials.
OpenCode has not disclosed the total number of affected accounts or confirmed the full extent of the attackers’ access. The 1.4% figure represents the share of Console users affected, not the percentage of exposed keys or compromised providers.
What the OpenCode Distillation Attack Means for API Security
A distillation attack involves querying an AI model extensively to collect outputs that can help train another model. Developers can use these responses to approximate aspects of a more capable model’s behavior.
In this incident, OpenCode identified a large-scale distillation attack but has not explained which models attackers targeted. It also has not disclosed the query volume, the duration of the activity, or the identity of those responsible.
Stolen API keys can allow attackers to make requests using a victim’s provider account. Depending on the provider’s billing arrangements and account permissions, unauthorized requests can generate charges for the key’s owner.
This creates a direct financial risk for developers and businesses that store provider credentials in external applications. A compromised key may remain usable until its owner revokes it or the provider otherwise disables access.
OpenCode instructed affected customers to revoke and replace exposed keys through their providers’ dashboards. Users must also update applications that rely on those credentials and review recent usage for suspicious activity.
Rotating keys quickly can prevent further unauthorized requests using the compromised credentials. Spending limits and usage alerts can help reduce potential losses, while secure credential storage can limit exposure.
The incident highlights the risks of linking external AI services to development platforms. Until OpenCode provides further findings, affected users should treat potentially exposed credentials as compromised and verify their provider usage.
