SUN, OCTOBER 18, 2026
Independent · In‑Depth · Practitioner‑Tested
✎ News

If You Use OpenCode, Rotate Your Keys Now

Anomaly Innovations, the company behind OpenCode, began emailing affected users late on 17 October about provider keys for OpenAI, Anthropic and Google that may have been exposed through a legacy console it was decommissioning. The earliest exposure it could identify was roughly 22 hours before that first notice.

By AIToolsRecap October 18, 2026 6 min read 38 views
Home › Articles › News › OpenCode Breach Exposed OpenAI and Anthropic AP...

If you have ever saved a provider API key in an OpenCode workspace, revoke and reissue it today. Anomaly Innovations, the company formerly known as SST that builds OpenCode, started emailing affected users at around 22:51 UTC on 17 October with the subject line "Action Required: Potential provider key breach." A second email followed just before 01:00 UTC on 18 October.

The intrusion came through a legacy console system the company was in the process of shutting down. Third-party provider keys saved in user workspaces — for OpenAI, Anthropic and Google — may have been exposed.

The timeline is the uncomfortable part

The earliest exposure the company could identify was approximately 00:30 UTC on 17 October. The first notification went out about 22 hours later. Twenty-two hours is a long time to hold a live API key that someone else may already have.

The detailed hour-by-hour reconstruction came from a developer posting as zephel01, not from the company. At the time of writing, OpenCode's status page, changelog and website carried no advisory at all — the only notice went by email, to accounts the company had already identified as affected. If you were not on that list, you were not told.

The company says it cannot confirm whether any keys were misused, and that its investigation to date found no evidence other data leaked. Two users have said otherwise: one reported unauthorised Claude API usage on the evening of 17 October, another roughly $100 in fraudulent charges.

What to do, in order

  • Revoke and reissue every affected key in the OpenAI, Anthropic and Google dashboards directly. Do not use links in the email.
  • Read your billing and usage logs back to 17 October. Unauthorised use shows up there before it shows up on an invoice.
  • Set spending limits on each provider account. If you had them already, this cost you $100 instead of $10,000.
  • Update every pipeline, CI job and local config that used the old keys, or you will be debugging 401s for a week.

The broader point: a coding agent that stores your provider keys is a credential store, whether or not it is marketed as one. This is the second credential-handling failure in the agent tooling space this month, after the Archon control API shipped with no authentication at all.

Sources: Startup Fortune

Tags
AI NewsOpenAIAnthropicGoogle2026Coding AI
⚑

Spot an inaccuracy?

We verify facts before publishing and correct errors promptly. If something in this article is wrong or outdated, let us know.

Report an error →