CVE-2026-18982: Odh-training-operator-rhel9: rhoai fork aggregates training job create onto native edit/admin clusterroles

Severity: high (CVSS 8.8 v3.1)
Exploited in the wild: not listed in CISA KEV
EPSS: 0.8% probability of exploitation in the next 30 days
Public exploit or detection template: none known
Published: 2026-08-10 · updated 2026-09-21
Weakness: CWE-250

Description

A flaw was found in the RHOAI training-operator. This vulnerability allows a user with standard edit or admin roles in any Kubernetes namespace to escalate their privileges. Through the creation of training jobs, an attacker can impersonate service accounts, access the host filesystem, and potentially execute arbitrary code remotely. This issue arises from the aggregation of training job permissions onto native Kubernetes edit and admin ClusterRoles, coupled with unrestricted PodTemplateSpec passthrough.

Affected products

Red Hat OpenShift AI 2.25: all versions — fixed in 1785187053
Red Hat OpenShift AI 3.3: all versions — fixed in 1785188461
Red Hat OpenShift AI 3.4: all versions — fixed in 1784814352, 1787361677
Red Hat OpenShift AI 3.5: all versions — fixed in 1786552250, 1786552271

How to fix

Upgrade to a fixed version: Red Hat OpenShift AI 2.25 1785187053; Red Hat OpenShift AI 3.3 1785188461; Red Hat OpenShift AI 3.4 1784814352, 1787361677; Red Hat OpenShift AI 3.5 1786552250, 1786552271. Check the vendor advisory for mitigations.

References

Get alerted when CVE-2026-18982 or the next vulnerability affects your stack