OpenCode Console breach: attackers ran up charges on users' stored AI provider keys

Why this label: The breach is confirmed by OpenCode. How attackers got in, how keys were stored, and the full extent of misuse had not been disclosed, and the incident was still described as ongoing when TechCrunch published. We read only the TechCrunch report; we did not see OpenCode's own statements directly and reproduced nothing.

Tried

Store your own OpenAI, Anthropic or Google API keys in OpenCode Console workspaces so the coding agent can use them.

Broke

Per TechCrunch (2026-10-19): OpenCode said attackers broke into an older system overnight, affecting 1.4% of OpenCode Console users, and users who did not get an email were not affected. Developers reported spend alerts firing across OpenAI, Anthropic and Google, with charges in the hundreds of dollars and a few totals above $1,000. The open-source app and CLI were not affected; the exposed keys were confined to Console, according to a co-founder.

Cause

• confirmed OpenCode confirmed a security breach of an older system tied to Console (its post on X, as reported by TechCrunch).
• suspected OpenCode called it a 'large-scale distillation attack'. TechCrunch notes some skeptics call that framing an overstretch and say it was a straightforward breach with the distillation use unproven.
• suspected How the attackers got in and how the keys were stored have not been disclosed, and the company said the incident remained ongoing when TechCrunch published.

“Confirmed” means the source states it as established; “suspected” means the source treats it as a theory or it is unsettled. We did not verify either ourselves.

Fix

OpenCode told affected users to revoke and replace their keys directly with each provider. Per TechCrunch, Anthropic and Google blocked the keys they believed were exposed, and OpenAI worked with OpenCode to restore access and refund its API customers for credits used in the attack; OpenCode said refunds for unauthorized charges were underway. Check your own provider dashboards for usage you did not make, and ask each provider's support about refunds.

Prevention

A pattern worth copying from one builder's cap guide (HelloBuilder, Oct 2026, not about this incident): give each project or agent its own provider key with its own spend limit, so a leaked key can only burn its own allowance, and set alerts at 50% and 80% of that limit. Keep keys out of third-party hosted stores when you can avoid it.

Sources

• TechCrunch (2026-10-19)
• HelloBuilder: Your Coding Agent Needs a Hard Budget Cap (2026-10-06)

← All stories