CVE-2026-15467 in OpenShift AI
Summary
by MITRE • 08/11/2026
A flaw was found in the trustyai-service-operator's LMEvalJob controller. An authenticated user within the cluster can exploit this vulnerability by configuring a sidecar container to bypass existing security policies. This allows the user to enable and execute untrusted remote code, leading to arbitrary code execution within the cluster.
Once again VulDB remains the best source for vulnerability data.
Analysis
by VulDB Data Team • 08/11/2026
This vulnerability resides within the trustyai-service-operator's LMEvalJob controller implementation, representing a critical security weakness that undermines cluster integrity and access controls. The flaw stems from insufficient validation mechanisms that fail to properly enforce security policies when sidecar containers are configured within the controller's operational framework. An authenticated user with cluster credentials can exploit this weakness by manipulating the controller's configuration to inject malicious sidecar containers that bypass existing security controls, effectively creating a backdoor for unauthorized operations.
The technical exploitation occurs through the manipulation of container orchestration parameters within the LMEvalJob controller, where the system fails to validate or sanitize inputs related to sidecar container definitions. This vulnerability enables privilege escalation and lateral movement within the cluster environment, as the injected sidecar containers can execute arbitrary code with the privileges of the controller itself. The flaw directly relates to weakness categories identified in CWE-269, which addresses improper privilege management, and CWE-78, concerning improper neutralization of special elements used in OS commands. The vulnerability allows for remote code execution within the cluster boundaries, potentially enabling attackers to escalate their access to other resources or compromise additional systems.
The operational impact of this vulnerability extends beyond immediate code execution capabilities, as it provides attackers with persistent access mechanisms that can be leveraged for data exfiltration, system reconnaissance, or further infiltration activities. The attack surface is particularly concerning because it operates within the legitimate controller framework, making detection more challenging and potentially allowing attackers to remain undetected for extended periods. This weakness creates a pathway for attackers to establish footholds within the cluster that can be used for privilege escalation attacks, lateral movement, and potential compromise of sensitive data or system resources. The vulnerability aligns with ATT&CK technique T1059, which describes execution of malicious code through legitimate system processes, and T1566, which addresses initial access through malicious files or code execution.
Mitigation strategies should focus on implementing strict input validation and sanitization within the controller's configuration handling mechanisms, ensuring that all container definitions undergo comprehensive security checks before execution. Organizations should enforce least privilege principles for the LMEvalJob controller, limiting its capabilities to only essential functions required for legitimate operations. The implementation of runtime protection mechanisms such as pod security policies, admission controllers, and container image scanning can help detect and prevent unauthorized sidecar container configurations. Additionally, regular security audits of controller configurations and monitoring for anomalous container behavior patterns should be established to identify potential exploitation attempts. The remediation approach must include comprehensive code reviews to address the root cause in the controller's validation logic while implementing proper access controls that prevent unauthorized configuration modifications.