The Data Science Pipelines Operator (DSPO) in Red Hat OpenShift AI has a ClusterRole with overly broad permissions, such as pods/exec and clusterrolebindings CRUD, which exceed its operational needs. This flaw allows an attacker who compromises the DSPO pod to escalate privileges to full cluster-admin level, enabling control over the entire Kubernetes cluster. The vulnerability is identified as CWE-250 (Execution with Unnecessary Privileges) and has a CVSS 3.1 base score of 8.7 (high severity). The vendor advisory recommends administrators reduce the ClusterRole permissions by removing unnecessary rights and limiting access to only essential resources. No official patch or fix is currently confirmed, so mitigation relies on manual permission adjustments and pod restarts.

If exploited, this vulnerability allows an attacker who gains access to the DSPO pod to escalate privileges to full administrative control over the Kubernetes cluster. This includes executing arbitrary commands within pods and managing cluster-wide roles and bindings, significantly increasing the blast radius of any initial compromise. The confidentiality and integrity of the cluster are highly impacted, while availability impact is not indicated.

No official patch or fix is currently confirmed for this vulnerability. Administrators should immediately review and modify the ClusterRole associated with the Data Science Pipelines Operator (DSPO) to remove unnecessary permissions such as pods/exec, kubeflow.org */*, seldondeployments *, and broad apiGroups:'*'. The ClusterRole should be restricted to only required resources like apps/deployments, services, secrets, configmaps, roles/rolebindings, routes, networkpolicies, servicemonitors, and DSPA/Argo CRDs. After applying these changes, restart the DSPO pod to ensure updated permissions take effect. Careful validation is necessary to avoid impacting operator functionality.