OpenCode breach may have exposed stored OpenAI, Anthropic and Google API keys

OpenCode, the open-source coding agent built by Anomaly Innovations, disclosed a security breach on October 17. The company says an intruder reached a legacy system on the night of October 16 and may have exposed third-party API keys that about 1.4% of OpenCode Console users had saved, the subset who linked external providers such as OpenAI, Anthropic and Google.

The company tied the incident to what it called a large-scale distillation attack, part of a mass effort to copy AI models, and its co-founder said attacks on its systems continued for 48 hours. OpenCode says the problem affects only the optional online Console service, not its desktop apps, which keep keys on the user's device. Affected users were notified by email, and the company says those who got no email are not affected. It has not confirmed whether any keys were copied or used, and as of October 19 it called the incident ongoing.

Some developers reported unauthorized charges, with one bill reported at $1,700. Anthropic said it revoked the Claude keys it believed were affected, Google said it blocked exposed Gemini keys, and OpenAI was reported to be refunding credits. Deleting a key inside Console does not cancel it, and budget limits cannot stop its use.

Our take

A stored API key is a bearer credential: whoever holds the string can spend your money. Any tool that keeps your provider keys on its servers is now part of your attack surface, whether you think of it that way or not.