OpenCode says a breach may have exposed API keys; users reported unauthorized AI charges.
Why it matters: Exposed developer credentials can create direct financial risk when AI coding platforms store keys for hosted model access. The reported losses remain user allegations, and OpenCode has not independently confirmed that keys were copied or used.
- OpenCode said 1.4% of Console users were affected and notified them on October 17, 2026.
- The potentially exposed credentials included keys for OpenAI, Anthropic and Google services.
- Users reported unauthorized charges of $281, $168 and approximately $1,700; the dossier does not identify provider or OpenCode verification of those amounts.
- OpenCode co-founder Dax Raad said the incident affected Console, not the local OpenCode desktop or CLI applications.
OpenCode said a security breach affected 1.4% of OpenCode Console users and that it notified affected customers by email on October 17, 2026. The company described the incident as a "large-scale distillation attack," but the available information does not establish that term as the mechanism of the credential exposure.
Console is OpenCode's hosted service, where users could save external provider credentials for hosted model requests. Potentially exposed keys included credentials for OpenAI, Anthropic and Google services. In comments reported by The Hacker News, co-founder Dax Raad said the incident affected Console rather than OpenCode's open-source desktop and command-line applications.
OpenCode's customer notice, as reported by The Hacker News, said attackers accessed a legacy system being decommissioned. The earliest possible exposure was October 17 at 00:30 UTC. OpenCode notified affected users that day, but the company has not publicly identified when it discovered the incident or when unauthorized access began.
Users reported unauthorized API spending, including $281 in OpenAI usage and $168 in Anthropic usage for one account. Another user reported approximately $1,700 in losses, according to TechCrunch. Those figures were user reports; the available information does not show that OpenCode or a provider independently verified them.
TechCrunch reported that Google suspended some projects, that Anthropic and Google blocked keys believed to be exposed, and that OpenAI worked with OpenCode to restore access and refund customers for credits used in the incident. OpenCode said the incident remained ongoing and that it was working with the three providers on remediation. Raad said attackers repeatedly attempted to access OpenCode systems for approximately 48 hours.
By the numbers
- 1.4% - share of OpenCode Console users the company said were affected
- 00:30 UTC on October 17, 2026 - earliest possible exposure identified in OpenCode's reported customer notice
- Approximately $1,700 - amount one user reported losing in unauthorized AI usage
Yes, but: OpenCode has not independently confirmed that the keys were copied or used, and the reported charges were not identified as verified by OpenCode or a provider.
What's next: OpenCode said the incident remained ongoing and that it was working with OpenAI, Anthropic and Google on remediation.