CVE-2026-18544 Overview
CVE-2026-18544 affects IBM Portieris versions 0.5.0 through 0.14.2. Portieris is a Kubernetes admission controller that enforces image trust policies before pods are admitted to a cluster. The vulnerability allows a remote authenticated attacker to bypass image policy enforcement due to improper authorization of pod owner references [CWE-862]. An attacker with cluster access can deploy container images that would otherwise be blocked by Portieris policies, undermining the trust guarantees the admission controller provides.
Critical Impact
Authenticated attackers can bypass image signature and policy enforcement, allowing untrusted or unsigned container images to run inside Kubernetes clusters.
Affected Products
- IBM Portieris 0.5.0 through 0.14.2
- Kubernetes clusters using Portieris as an admission controller
- Workloads relying on Portieris image trust policies for compliance
Discovery Timeline
- 2026-08-19 - CVE-2026-18544 published to NVD
- 2026-08-20 - Last updated in NVD database
Technical Details for CVE-2026-18544
Vulnerability Analysis
IBM Portieris intercepts pod creation requests and validates container images against configured ImagePolicy and ClusterImagePolicy resources. When a pod is created by a higher-level controller such as a Deployment, ReplicaSet, or Job, Portieris uses the pod's ownerReferences field to determine which policy scope applies. The controller does not sufficiently verify that the referenced owner actually exists or that the requesting user has authority over that owner. An authenticated attacker can craft a pod manifest containing fabricated ownerReferences that point to workloads in namespaces with permissive policies. Portieris then evaluates the request under the attacker-supplied context rather than the true creation context, permitting images that violate the enforced policy.
Root Cause
The root cause is a missing authorization check [CWE-862] on pod owner references. Portieris trusts client-supplied ownerReferences metadata during policy resolution without confirming ownership or the caller's permissions on the referenced object.
Attack Vector
Exploitation requires network access to the Kubernetes API server and valid credentials with permission to create pods in at least one namespace. The attacker submits a pod specification with ownerReferences designed to steer policy resolution toward a permissive scope. See the IBM Support Article for vendor technical details.
Detection Methods for CVE-2026-18544
Indicators of Compromise
- Pod manifests containing ownerReferences that reference objects the creating user does not own or cannot access.
- Running pods with container images that do not match signed digests defined in ImagePolicy or ClusterImagePolicy resources.
- Portieris admission webhook logs showing policy resolution against namespaces or owners inconsistent with the request source.
Detection Strategies
- Audit Kubernetes API server logs for CREATE pods requests where ownerReferences point to non-existent or cross-namespace objects.
- Compare deployed container image digests against the intended signed image list and flag unsigned or unexpected images.
- Alert on discrepancies between the requesting ServiceAccount and the owner objects declared in pod metadata.
Monitoring Recommendations
- Enable full Kubernetes audit logging at the RequestResponse level for pod creation events.
- Forward Portieris admission controller logs to a centralized SIEM for correlation with API server activity.
- Monitor for sudden increases in pod creations that bypass expected policy scopes or use unusual owner references.
How to Mitigate CVE-2026-18544
Immediate Actions Required
- Upgrade IBM Portieris to a release later than 0.14.2 that contains the fix referenced in the IBM Support Article.
- Review and restrict pods/create RBAC permissions to minimize the population of users able to exploit the flaw.
- Audit current cluster workloads for images that do not match approved signatures and remove any unauthorized deployments.
Patch Information
IBM has published remediation guidance in the IBM Support Article. Administrators should apply the fixed Portieris version and redeploy the admission webhook across all clusters.
Workarounds
- Enforce image trust at an additional layer, such as Sigstore policy-controller or Kyverno, so a single Portieris bypass does not permit untrusted images.
- Apply Kubernetes ValidatingAdmissionPolicy rules that reject pods whose ownerReferences cannot be resolved to existing objects.
- Tighten RBAC so only trusted controllers and service accounts can create pods directly, forcing workloads through higher-level objects.
# Verify installed Portieris version and upgrade via Helm
kubectl -n ibm-system get deployment portieris -o jsonpath='{.spec.template.spec.containers[0].image}'
helm repo update
helm upgrade portieris ibm/portieris -n ibm-system --version <fixed-version>
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.

