CVE-2026-18951

Awaiting Analysis Awaiting Analysis - Queue

Privilege Escalation in Red Hat OpenShift AI TrainJobs

Publication date: 2026-08-10

Last updated on: 2026-08-27

Assigner: redhat-SADP

Description

A flaw was found in the Red Hat OpenShift AI (RHOAI) overlay for the training operator. The RHOAI overlay incorrectly aggregates `trainjobs` management permissions into the native Kubernetes `edit ClusterRole`. This allows any user with `edit ClusterRole` permissions in a namespace to create, modify, and delete `TrainJobs`. When combined with a separate vulnerability (TRN-01) that permits arbitrary pod configurations, a remote attacker with namespace editor privileges could exploit this to escalate privileges, potentially leading to arbitrary code execution.

CVSS Scores

EPSS Scores

Probability:
Percentile:

Meta Information

Published

2026-08-10

Last Modified

2026-08-27

Generated

2026-08-31

AI Q&A

2026-08-11

EPSS Evaluated

2026-08-29

NVD

CVE-2026-18951

Affected Vendors & Products

Vendor	Product	Version / Range
red_hat	openshift_ai	*

Helpful Resources

Exploitability

CWE

KEV

CWE ID	Description
CWE-284	The product does not restrict or incorrectly restricts access to a resource from an unauthorized actor.

Attack-Flow Graph

Executive Summary

This vulnerability is in Red Hat OpenShift AI (RHOAI) where the training operator's permissions are incorrectly set. The native Kubernetes edit ClusterRole is given elevated permissions to manage TrainJobs, allowing users with edit permissions to create, modify, and delete TrainJobs. Combined with another vulnerability (TRN-01), this could let an attacker with namespace editor privileges escalate privileges and execute arbitrary code.

Detection Guidance

To detect this vulnerability, check if users with the native Kubernetes edit ClusterRole have permissions to create, modify, or delete TrainJobs in your OpenShift AI environment. Review RoleBindings and ClusterRoleBindings for the edit ClusterRole in each namespace. Look for unexpected TrainJob resources or modifications in namespaces where users have edit permissions.

Impact Analysis

If you have edit ClusterRole permissions in a namespace, an attacker could exploit this to gain higher privileges. This may allow them to run unauthorized code, access sensitive data, or disrupt services in your environment.

Compliance Impact

This vulnerability could potentially violate compliance with GDPR and HIPAA by allowing unauthorized privilege escalation and arbitrary code execution. Unauthorized access to training jobs and pod configurations may lead to data breaches or unauthorized modifications, compromising data integrity and confidentiality required by these regulations.

Mitigation Strategies

Review and restrict permissions for users with the edit ClusterRole in namespaces where Red Hat OpenShift AI is deployed. Remove TrainJob management permissions from the edit ClusterRole to prevent unauthorized creation, modification, or deletion of TrainJobs.

Hi! I’m here to help you understand CVE-2026-18951. Ask me anything about the vulnerability, its impact, or mitigation strategies.

0/70