OpenCode says an intrusion on the night of 16 October 2026 hit 1.4% of OpenCode Console users. Provider keys saved in those workspaces may be exposed. If you store OpenAI, Anthropic or Google keys in any third-party tool, revoke the ones linked to Console and issue new ones.
The short answer
- Who: 1.4% of Console users, each emailed by OpenCode. No email means not affected, per OpenCode.
- What: provider keys, such as OpenAI and Anthropic keys, saved in a workspace may be exposed.
- Best move: revoke linked keys at each provider and create new ones. Removing a key from Console does not revoke it.
- Still open as of 19 October: the cause, the amount of misuse, the attacker’s identity and any compensation policy.
We could not open OpenCode’s original statement, so the wording below is as quoted by Infosecurity Magazine, IBTimes Singapore and StepSecurity.
What does OpenCode say is in scope?
| Question | What is attributed to OpenCode |
|---|---|
| Who was affected? | 1.4% of Console users; each was emailed, and “if you did not get an email you were not impacted” |
| What may be exposed? | Provider keys, such as OpenAI and Anthropic keys, saved in a workspace, per the user notice quoted by Infosecurity |
| How did it happen? | Unauthorized access to a “legacy system” the company was decommissioning, per the same notice; the company links the incident to a “large scale distillation attack” |
| Is it over? | Described as ongoing, with the company working with OpenAI, Anthropic and Google, per StepSecurity and IBTimes |
| More detail? | StepSecurity reports the company said it cannot share more at this point |
Does it affect local installs?
Co-founder Dax Raad is quoted by IBTimes (17 October) saying this is the Console, “not OpenCode itself,” and that those keys “never come to us.” IBTimes adds that keys added in the CLI and desktop apps are stored locally. That detail is the outlet’s, not a quote from OpenCode.
So a CLI-only user is probably not in the affected group. We suggest checking your inbox and your provider usage anyway.
What is the timeline as reported?
| Date | Reported |
|---|---|
| Night of 16 October | The intrusion, per OpenCode; no time given |
| 17 October | OpenCode’s post on X and emails to affected users “this morning,” per Infosecurity and IBTimes |
| 18 October | StepSecurity’s post |
| 19 October | Infosecurity and IBTimes coverage |
What do third parties say?
- Anthropic: IBTimes quotes Lydia Hallie saying the company worked with OpenCode and revoked Claude API keys believed to be affected. She advised creating a new key to be extra careful.
- Users: IBTimes cites individual users who report unexpected charges at providers. A link to this incident is unconfirmed, so do not add those reports up into a total loss.
- The distillation-attack link: it appears only in OpenCode’s own statement. Neither outlet reports independent verification or an attacker.
- Providers: nothing we read says the providers’ own systems were breached.
- StepSecurity: it sells security products and relies on OpenCode’s statement. It is not an independent investigation of the cause.
What should you do now?
These are the steps the coverage recommends, not legal or financial advice.
- Revoke any OpenAI, Anthropic or Google key linked to Console at each provider’s dashboard and create a new one. IBTimes notes that removing a key from Console does not revoke it at the provider.
- Review each provider’s usage for requests you did not make, from 16 October onward (guidance quoted by Infosecurity).
- Prefer restricted replacement keys with spending limits, as the same guidance suggests.
- We suggest updating keys in CI/CD and environment files, then confirming the revoked keys fail authentication.
- Search your inbox for the subject “Action Required: Potential provider key breach,” as Infosecurity suggests.
What is still unknown?
- Root cause, how the legacy system was reached and for how long.
- The absolute number of users or keys affected, and whether keys were actually copied or used.
- How the “distillation attack” relates to the intrusion, and who was behind it.
- Any published compensation policy, and whether anything besides provider keys was exposed (StepSecurity says this is not disclosed).
Questions and answers
Were OpenAI, Anthropic or Google breached?
Nothing in the coverage says so. The reports concern keys saved by accounts in OpenCode Console.
Am I affected if I only use the CLI?
Probably not. Per IBTimes’ account of OpenCode’s comments, the incident concerns the Console rather than OpenCode itself, and CLI keys are stored locally.
How do I know if I was notified?
OpenCode says affected users were emailed and that no email means not affected. Search your inbox for the subject line above, and review your provider usage either way.
Is deleting the key from Console enough?
No. IBTimes notes that removing a key from Console does not revoke it at the provider. Revoke it in the provider’s dashboard.
Will I be reimbursed for misuse?
No compensation policy is published in the coverage we read. Document any unexpected charges and ask the provider directly.
Sources
- Infosecurity Magazine: OpenCode breach
- IBTimes Singapore: OpenCode breach
- StepSecurity: OpenCode security breach (the publisher sells security products)
Status as of 19 October 2026. Quotes are as relayed by the coverage above. Details and remediation status may change, so check OpenCode’s official notices.
