Skip to content
Threat Feed
Subscribe
high advisory

CVE-2026-16745: Authentication Bypass in Red Hat OpenShift AI odh-dashboard

A critical vulnerability, CVE-2026-16745, exists in the odh-dashboard web console component of Red Hat OpenShift AI (RHOAI), allowing a malicious actor within the cluster to bypass authentication by providing an arbitrary access token, leading to user impersonation and unauthorized access to the Kubernetes API, potentially resulting in arbitrary code execution, privilege escalation, and information disclosure.

A critical vulnerability, CVE-2026-16745, has been discovered in odh-dashboard, the web console component of Red Hat OpenShift AI (RHOAI). This flaw, stemming from incorrect network binding, allows a malicious actor who has already gained access to the cluster to bypass authentication mechanisms. By presenting an arbitrary access token, the attacker can impersonate any user within the cluster, thereby gaining unauthorized access to the Kubernetes API. This exploitation can lead to severe consequences, including arbitrary code execution, privilege escalation, and sensitive information disclosure across the OpenShift AI environment. This vulnerability poses a significant risk to the integrity and confidentiality of data and operations within affected Red Hat OpenShift AI deployments, making immediate mitigation crucial for defenders.

Attack Chain

  1. An adversary gains initial access to the Red Hat OpenShift AI (RHOAI) cluster via an unspecified method (pre-condition for this CVE).
  2. The adversary identifies the odh-dashboard web console component as a target for privilege escalation or lateral movement.
  3. The adversary crafts a request exploiting an incorrect network binding flaw within odh-dashboard.
  4. By providing an arbitrary access token to the vulnerable component, the adversary successfully bypasses odh-dashboard's authentication mechanisms.
  5. The adversary successfully impersonates any legitimate user within the cluster, leveraging their arbitrary access.
  6. Using the impersonated user's identity, the adversary gains unauthorized access to the Kubernetes API.
  7. Through the compromised Kubernetes API access, the adversary executes arbitrary code, achieves higher privilege escalation, or exfiltrates sensitive information.

Impact

Successful exploitation of CVE-2026-16745 leads to a complete bypass of authentication within the odh-dashboard component, allowing an adversary to impersonate any user. This direct access to the Kubernetes API with elevated privileges enables arbitrary code execution across the cluster, significant privilege escalation to control critical cluster resources, and extensive information disclosure of sensitive data. Organizations using Red Hat OpenShift AI are at risk of full cluster compromise, data theft, and disruption of AI workloads if this vulnerability is exploited by an internal or otherwise pre-authenticated malicious actor.

Recommendation

  • Prioritize patching Red Hat OpenShift AI to address CVE-2026-16745 immediately as per the Red Hat security advisories listed in the references.
  • Implement robust monitoring of Kubernetes API audit logs for unusual access patterns, especially those originating from within the cluster or demonstrating user impersonation attempts.
  • Monitor for process creation events and network connections related to odh-dashboard or the Kubernetes API that deviate from normal baseline behavior, focusing on any attempts to provide arbitrary access tokens or unexpected privilege changes.
  • Review access controls and network segmentation within the cluster to limit the blast radius of a compromised internal actor.