RC RANDOM CHAOS

OpenCode Console Breach Exposes Stored AI Provider Keys

· via The Hacker News

Original source

OpenCode Console Breach May Have Exposed OpenAI, Anthropic, and Google API Keys

The Hacker News →

A breach of OpenCode’s hosted Console service exposed API keys for OpenAI, Anthropic, and Google stored by 1.4% of users. An intruder accessed a legacy system on October 16, prompting OpenCode to notify affected users the next day and urge them to revoke and replace their credentials.

API keys function as bearer tokens that enable billable requests against an account without additional verification. Deleting a key from Console does not revoke it at the provider level. Several users report unauthorized charges ranging from $168 to $1,700, though OpenCode has not confirmed whether the exposed keys were actually copied. The incident appears connected to a broader distillation campaign that harvests model outputs for training rival systems.

OpenCode emphasized the compromise was limited to its hosted Console; keys stored locally in its command-line and desktop applications remained secure. Affected users should revoke keys at each provider, rotate all applications using the old credentials, audit charges from October 16 forward, and implement per-tool API key isolation with spending limits. Attackers continued attempting to breach OpenCode’s systems through at least October 18, driving accelerated security hardening.

Read the full article

Continue reading at The Hacker News →

This is an AI-generated summary. Read the original for the full story.