OpenShift AI Flaw Allows Full Cluster Takeover (CVE-2025-10725)
A severe vulnerability in Red Hat OpenShift AI (CVE-2025-10725) allows low-privileged users to escalate their permissions to full cluster administrator. The flaw stems from an overly permissive ClusterRole assignment that grants any authenticated user the ability to create jobs across the entire cluster. An attacker with minimal access, such as a data scientist account, can abuse this permission to run malicious jobs with elevated rights. This could lead to a complete compromise of the cluster, including data theft and service disruption. Administrators are urged to mitigate the risk by immediately removing the insecure ClusterRoleBinding and adopting a least-privilege security model.
Latest mentioned: 10-01
Earliest mentioned: 10-01