Hackers breach OpenCode, charging users' AI accounts
OpenCode, an open source AI coding agent, confirmed attackers broke into an older system overnight, compromising API keys for 1.4% of its Console platform users. Affected users reported unauthorized charges ranging from hundreds to over a thousand dollars across OpenAI, Anthropic, and Google accounts. Anthropic and Google blocked exposed keys; OpenAI refunded affected customers. The open source app itself was not affected. The incident also exposed a broader defensive gap: AI models refused to analyze malicious code during the response because safety filters blocked the work, until OpenCode gained access to OpenAI's Daybreak Blue program for vetted security users.
