CVE-2026-16745 is a vulnerability in the odh-dashboard component of Red Hat OpenShift AI that arises from incorrect network binding. This flaw permits an attacker with access inside the cluster to bypass authentication by submitting an arbitrary access token, effectively impersonating any user. Exploitation can grant unauthorized access to the Kubernetes API, potentially resulting in arbitrary code execution, privilege escalation, or data disclosure. The vulnerability is classified under CWE-346 (Origin Validation Error) and has a CVSS 3.1 score of 8.8 (high severity). Red Hat has published security advisories RHSA-2026:53261 and RHSA-2026:53262, which include updated RHOAI images (2.25.10 and 3.4.3) that mitigate this issue. Detailed upgrade instructions are available in Red Hat's official documentation.

An attacker with access inside the cluster can bypass authentication controls in the odh-dashboard component, impersonate any user, and gain unauthorized access to the Kubernetes API. This can lead to critical impacts including arbitrary code execution, privilege escalation, and information disclosure within the affected environment. The vulnerability is rated high severity with a CVSS score of 8.8, reflecting its potential to compromise confidentiality, integrity, and availability.

Red Hat has released updated versions of Red Hat OpenShift AI (RHOAI) that address this vulnerability: versions 2.25.10 and 3.4.3. Users should upgrade their RHOAI deployments to these versions following the official Red Hat upgrade documentation to fully mitigate the issue. No other temporary or workaround mitigations are specified in the vendor advisory. Patch status is confirmed as fixed in these releases.