The vulnerability CVE-2026-6859 affects Red Hat Enterprise Linux AI (RHEL AI) 3 through the InstructLab component. The linux_train.py script sets trust_remote_code=True when loading models from the HuggingFace Hub, which implicitly trusts remote code execution. An attacker can exploit this by supplying a malicious model that, when loaded by the user running ilab train/download/generate, executes arbitrary Python code. This can result in full system compromise. The CVSS v3.1 base score is 8.8, indicating high severity with network attack vector, low attack complexity, no privileges required, user interaction required, and high impact on confidentiality, integrity, and availability. No patch or official remediation level is provided in the vendor advisory content.

Successful exploitation allows remote attackers to execute arbitrary Python code on the affected system, potentially leading to complete system compromise. The vulnerability impacts confidentiality, integrity, and availability of the system. Since the attack requires user interaction (running a specific command with a malicious model), the risk depends on user behavior. There are no known exploits in the wild at this time.

Patch status is not yet confirmed — check the Red Hat advisory at https://access.redhat.com/security/cve/CVE-2026-6859 for current remediation guidance. Until an official fix is available, users should avoid running ilab train/download/generate commands with untrusted or unverified models from the HuggingFace Hub. Review and restrict usage of models from external sources to trusted repositories only.