OpenCode Breach Exposes API Keys in AI Model Distillation Attack
OpenCode users who linked external AI services suffered credential exposure in an attack designed to run unauthorized queries at scale. Attackers conducted a distillation attack—training cheaper, smaller models by extracting knowledge from Anthropic, OpenAI, and Google’s services—affecting roughly 1.4% of the platform’s user base. The impact was swift and costly: some users incurred over $1,700 in fraudulent charges overnight, while Google suspended projects displaying signs of compromised resources.
Assailants probed OpenCode’s systems using AI agents for at least two days before discovery. When the platform’s own security filters began blocking suspicious queries, investigation stalled until OpenCode deployed OpenAI’s Daybreak cybersecurity model for defense, which stabilized the situation. Credentials were exposed starting October 17, 2026, UTC early morning. OpenCode is coordinating with affected providers and indicated possible reimbursement for users hit with unauthorized charges.
Read the full article
Continue reading at BleepingComputer →This is an AI-generated summary. Read the original for the full story.