Northflank breach: customers told to assume all stored secrets were decrypted and rotate them, AI API keys reported used
Northflank disclosed a security incident on October 30, 2026 and told every customer to assume their environment variables had been decrypted and rotate all stored secrets immediately, after users reported unauthorised use of AI API keys stored on the platform; the company has given no root cause and no affected count.
What happened
Northflank, a cloud platform that lets developers deploy and manage applications, containers, databases and AI applications without running the underlying infrastructure, disclosed a security incident on its status page on October 30, 2026. The timeline Northflank published starts with a threat alert from cloud security company Wiz at 11:43 UTC that day: its engineers reviewed the alert a minute later, deployed network policy patches at 12:02, 12:50 and 13:05 UTC, and requested support from Wiz’s incident response team. At 13:40 UTC Northflank told customers to “Rotate environment variables immediately”, and at 13:58 UTC the instruction hardened to: “Assume all environment variables have been decrypted. Disable/rotate all keys.” Later on October 30 the company relayed that multiple users were confirming their AI keys had been used, and at 23:38 UTC it stated in writing that “Users have reported unauthorised use of AI API keys supplied to the Northflank platform.” On October 31 it enabled audit logs for all customers, including free users, and expanded its built-in secret rotation to Postgres, MongoDB, MySQL, Redis, Minio and RabbitMQ add-ons. Northflank says its investigation is ongoing and that it has “no further confirmed findings to share at this time” — no root cause, no attacker, and no affected count has been published. This report is labelled POTENTIAL EXPOSURE because the company instructs every customer to assume compromise but has not confirmed how many customers’ secrets were actually taken.
Who is affected
Every Northflank customer who stored secrets on the platform — Northflank’s guidance is platform-wide, not limited to a named subset, and it has published no count of affected customers. The secrets in scope are the user-provided environment variables held in workloads, in secret groups and in global secrets: the values applications use to reach outside services, which commonly include AI provider API keys, database passwords, cloud service credentials and authentication tokens. The only misuse Northflank has specifically reported is of AI API keys supplied to the platform, raised by the users themselves. The company has not said whether anything beyond stored secrets was accessed, whether customer application data was touched, or how the environment variables came to be at risk.
How serious
The concrete risk is that a stored secret can be spent or used the moment someone else holds it, with no further break-in needed. An AI API key is billed by use, so unauthorised use — which Northflank customers have already reported — turns directly into charges on the key owner’s account and can run until the key is revoked. The other classes of secret Northflank holds are the keys to connected services: a database password, a cloud credential or an authentication token taken from an environment variable can open the service it belongs to, so one platform incident can cascade into accounts the attacker never touched directly. Northflank’s own instruction — assume every environment variable was decrypted — also means a customer who has seen no suspicious activity cannot treat that silence as safety: the company is telling all of them to act as if their secrets are already out.
What to do now
Follow Northflank’s guidance, in its order. Revoke existing credentials with the providers that issued them first — changing a value only inside Northflank leaves the old key working for whoever holds it — then generate replacements and update the affected secrets and workloads in Northflank. Rotate all user-provided secrets stored in workloads, secret groups and global secrets, not only AI keys, using the rotation Northflank has enabled for its Postgres, MongoDB, MySQL, Redis, Minio and RabbitMQ add-ons where those apply. Review your AI provider usage and billing logs for unexpected activity, and report anything suspicious to Northflank support and to the provider concerned. Disable public access to add-ons where you have enabled it. Finally, read the audit logs Northflank switched on for all customers on October 31, and check the logs of the services your keys reach, for changes, logins or deployments you cannot account for.
This is the same pattern as the Zeabur breach we reported on October 20, 2026 — see our Zeabur report — where stored environment variables and AI service keys were the prize. Wiz, whose alert started Northflank’s response, counted Northflank among nine AI developer platforms that disclosed incidents in October 2026 in its “Black October” research.
Sources: Northflank status page, “Security incident” (updates October 30–31, 2026) · BleepingComputer, October 31, 2026