Red Hat Openshift AI Service Vulnerability Allow Attackers to Take Control of the Infrastructure
Red Hat has issued a security advisory, CVE-2025-10725, identifying an important vulnerability in the OpenShift AI Service. This flaw allows low-privileged users to escalate their permissions to full cluster administrator, potentially compromising the…
Red Hat has issued a security advisory, CVE-2025-10725, identifying an important vulnerability in the OpenShift AI Service. This flaw allows low-privileged users to escalate their permissions to full cluster administrator, potentially compromising the entire platform.
The vulnerability carries a CVSS v3 base score of 9.9, indicating a critical risk for organizations utilizing Red Hat OpenShift AI for machine learning tasks.
Privilege Escalation Vulnerability (CVE-2025-10725)
The vulnerability originates from an overly permissive ClusterRoleBinding, which connects the kueue-batch-user-role with the system:authenticated group.
In standard deployments, authenticated users, such as data scientists, typically access Jupyter notebooks and AI pipeline features with limited privileges.
By exploiting this configuration, an attacker can use the batch.kueue.openshift.io API to create arbitrary Job and Pod resources.
This allows execution of malicious containers or init-containers that perform oc or kubectl commands, facilitating privilege elevation within the cluster by binding newly created service accounts to roles with higher privileges.
This escalation can lead to the attacker assuming the cluster-admin role, granting unrestricted access to all cluster objects, which could result in data theft, service disruption, and full infrastructure control.
Red Hat has issued a security advisory, CVE-2025-10725, identifying an important vulnerability in the OpenShift AI Service.
Risk Factors Details
Affected Products Red Hat OpenShift AI 2.19 (RHEL 8), Red Hat OpenShift AI 2.21 (RHEL 9), registry.redhat.io/rhoai/odh-rhel8-operator, registry.redhat.io/rhoai/odh-rhel9-operator
Impact Privilege escalation
Exploit Prerequisites Valid authenticated user account, Access to OpenShift AI Service, Low-privileged account, Access to Jupyter notebook or similar interface, Ability to interact with batch.kueue.openshift.io API
CVSS 3.1 Score 9.9 (Critical)
Red Hat has released fixes through the following errata, updating the OpenShift AI Operator for Red Hat Enterprise Linux 8 and 9:
RHBA-2025:16984 (OpenShift AI 2.19 on RHEL 8) RHBA-2025:16983 (OpenShift AI 2.21 on RHEL 9)
Administrators are advised to apply these updates immediately. As an interim measure, prior to patching, the problematic ClusterRoleBinding should be removed.
Subsequently, job-creation permissions should be explicitly granted to trusted identities, enforcing the Principle of Least Privilege, ensuring only authorized users or groups can submit AI jobs.
Red Hat categorizes CVE-2025-10725 as Important, rather than Critical, due to the requirement of a valid authenticated account, albeit with low privileges.
However, the potential impact remains significant. Organizations utilizing Red Hat OpenShift AI are urged to address this vulnerability promptly to prevent a complete cluster compromise and to maintain stringent RBAC configurations.
Based on reporting by Cyber Security News.
