OpenCode breached in 'distillation attack'; users' OpenAI and Anthropic balances drained

OpenCode, the open-source coding agent, was hacked, exposing OpenAI, Anthropic and Google API keys that some users had linked to their accounts.

The company said on X on Oct. 17 that a security breach "related to a large scale distillation attack" hit 1.4% of OpenCode Console users the night before. Only a subset of users with external providers linked were affected, and all of them were notified by email. "The incident is ongoing so we cannot share more details," OpenCode said, adding that it is working with Anthropic, OpenAI and Google. The post passed 510,000 views.

According to Startup Fortune, the keys leaked through a legacy console system that parent company Anomaly Innovations, formerly SST, was in the process of shutting down.

Victims quickly surfaced. "My OpenAI and Anthropic balances got drained, Google account got limited," wrote Max Weinbach, who put his loss at about $450 in credits plus $16 he had paid for. Developer Chris Sinco said spend alerts from all three providers fired before he went to bed. Security researcher Matt Johansen's post — "Confirmed. OpenCode hacked. Keys stolen." — drew 1,900 likes.

Anthropic's Lydia Hallie said the company had revoked Claude API keys it believed were affected and urged anyone who saved a key in OpenCode to create a new one. OpenCode's Dax wrote that his personal accounts were now under attack too: "whole new world with ai powered hacking."