OpenCode says hackers accessed an old system connected to its hosted Console, potentially exposing AI-provider keys for 1.4% of Console users and leading to unauthorized charges, with refunds underway.
OpenCode said affected users had connected outside AI providers and saved their API keys in Console workspaces. It notified them by email. It said users who received no notice were not affected.
OpenCode described the incident as related to a large-scale “distillation attack.” But it has not publicly explained how the attackers got into the system. It also has not shown how the breach enabled model distillation.
Users reported unexpected bills from AI services including OpenAI, Anthropic and Google. Some charges reached hundreds of dollars. A few reports topped $1,000.
Anthropic and Google blocked keys they believed had been exposed. OpenAI worked with OpenCode to restore affected users’ access. It also worked with OpenCode to refund credits used in the attack.
OpenCode said refunds were underway. The company emphasized that the breach involved Console, not its open-source coding app or its tool for entering commands in a terminal. Affected users were urged to disable their keys and get new ones from their AI providers.
They were also urged to check their accounts for use they did not recognize.