The Northflank security incident has prompted an urgent credential-rotation warning for customers of the cloud application hosting platform. Northflank says users reported unauthorized use of AI API keys stored on its service. On October 31, 2026, the company expanded its secret-rotation tools to additional database and messaging add-ons while continuing to investigate the October 30 incident.
What happened at Northflank?
Northflank provides infrastructure for deploying applications, databases and AI workloads. According to the company’s public incident timeline, cloud security firm Wiz alerted Northflank at 11:43 UTC on October 30. Northflank engineers reviewed the alert, applied an initial network-policy patch at 12:02 UTC and deployed further changes later that day. The company also requested assistance from Wiz’s incident-response team.
Northflank then instructed customers to treat environment variables as potentially exposed and immediately rotate secrets stored in workloads, secret groups and global secrets. Environment variables often contain passwords, API tokens and service credentials used by running applications.
What is confirmed as of November 1?
Northflank’s status updates say multiple customers reported unauthorized use of AI API keys supplied to the platform. The company specifically advised users to revoke affected credentials at the issuing providers, create replacements, update their Northflank workloads and review AI-service usage and billing records for suspicious activity.
The company enabled audit logs for all customers, including free accounts, on October 31 at 01:04 UTC. Credential rotation for PostgreSQL and MongoDB had been enabled on October 30. New October 31 updates made rotation available for MySQL and Redis at 13:34 UTC, MinIO at 22:32 UTC and RabbitMQ at 23:31 UTC. Customers should use the current platform guidance for each service and verify that replacement credentials are active.
As of the latest published incident update reviewed for this article, Northflank’s status page still labeled the security incident as identified rather than resolved. The status page showed major-outage labels for several platform components; those labels alone do not establish that every customer workload was unavailable.
What remains unverified?
Northflank has not publicly established the initial intrusion method, the number of affected customers, the full extent of credential access or the identity of the attacker. Its instruction to assume all environment variables were decrypted is precautionary incident-response guidance, not a published finding that every customer’s secrets were accessed.
Northflank noted similarities to activity described by Wiz in a broader report about attacks on AI development platforms. Wiz documented incidents at multiple providers and cases of stolen AI-service credentials being abused, but did not establish that every incident involved the same attackers. The reported abuse in Wiz’s other cases should not be attributed to Northflank without evidence.
Who is affected and why does it matter to businesses?
Organizations that stored application secrets, database passwords, cloud credentials or AI-provider keys in Northflank should assess their exposure. The immediate risk is not limited to the hosting platform: a credential issued by another service can sometimes be used outside Northflank until that provider revokes it.
For small businesses using hosted applications or AI tools, stolen API keys may enable unauthorized model requests, unexpected charges or access to connected systems, depending on the permissions assigned to each key. The incident is a reminder to review what third-party platforms can access and how quickly credentials can be replaced.
What Northflank customers should do now
Revoke and replace stored secrets
Inventory every credential stored in Northflank workloads, secret groups, global secrets and add-ons. Revoke the old credential at its issuing service, generate a new one with the minimum necessary permissions, update the workload and confirm the application still works. Merely changing an environment-variable value does not invalidate a previously copied token.
Check AI usage and cloud billing
Review API-call histories, spend, model usage and access locations at AI providers and other connected services. Investigate unfamiliar activity, apply usage alerts or spending limits where available and report suspected abuse through the provider’s official support channel.
Review logs and limit public exposure
Use the newly available Northflank audit logs and provider-side logs to examine unexpected access or configuration changes. Disable public access to add-ons where it is unnecessary, as Northflank advises. Retain relevant logs for investigation and consider additional downstream credential or session rotation if exposed keys could grant broader access.
Prepare for a vendor credential incident
Keep a record of which vendors hold sensitive credentials, identify owners for each key and maintain a tested rotation procedure. Separate production and test credentials, use narrowly scoped service accounts and avoid reusing a secret across unrelated systems.
Where to find updates
Northflank’s incident status page remains the primary source for new findings and remediation instructions. The company’s investigation is ongoing, and the affected-customer count and root cause may change as more evidence becomes available.