A critical authentication bypass vulnerability in Red Hat OpenShift AI (RHOAI) allows unauthorized actors to forge headers and gain elevated access within the cluster.

This is an authentication bypass vulnerability (CWE-290) occurring in the MaaS API. The system incorrectly trusts user-provided X-MaaS-Username and X-MaaS-Group headers without verification, allowing an authenticated pod to impersonate other users or groups.

The vulnerability carries a CVSS score of 9.9, reflecting its critical potential for systemic compromise. Successful exploitation enables an attacker to mint Kubernetes ServiceAccount tokens, revoke API keys, and exfiltrate sensitive configuration data, resulting in a total loss of confidentiality and integrity for the affected AI model environment.

Immediate Action: Update Red Hat OpenShift AI (RHOAI) to the latest available version as provided by Red Hat. Consult the vendor advisory at https://access.redhat.com/security/cve/CVE-2026-14450 for specific release information.

Proactive Monitoring: Monitor cluster logs for anomalous HTTP header patterns or unexpected service account token generation requests. Audit access logs for unauthorized attempts to access cross-tenant resources.

Compensating Controls: Implement strict network policies to isolate pods and limit communication between tenants until the patch is applied.

Public Exploit Available: No

Given the critical CVSS severity and the nature of the vulnerability allowing full privilege escalation across tenant boundaries, organizations must prioritize patching this component immediately. Failure to address this flaw could lead to complete exposure of sensitive AI infrastructure and data.