CVE-2026-31230 is a critical vulnerability in the Adversarial Robustness Toolbox (ART) through version 1.20.1. The Kubeflow component script robustness_evaluation_fgsm_pytorch.py uses the unsafe eval() function to parse string values passed via the --clip_values and --input_shape command-line arguments. This unsafe use of eval() enables an attacker who can control these arguments to inject arbitrary Python code, which will be executed on the system running the ART evaluation. The vulnerability can be exploited remotely if the attacker has the ability to influence these command-line arguments, potentially through pipeline configuration or automated scripts. The CVSS v3.1 base score is 9.8, indicating critical severity with network attack vector, low attack complexity, no privileges or user interaction required, and high impact on confidentiality, integrity, and availability. No official patch or remediation level is currently stated in the vendor advisory, and no known exploits in the wild have been reported. The vendor advisory from Red Hat is available but does not specify a fix or mitigation.

Successful exploitation allows remote attackers to execute arbitrary Python code on the system running the ART Kubeflow evaluation script, potentially leading to full system compromise. The vulnerability impacts confidentiality, integrity, and availability with high severity as indicated by the CVSS score of 9.8. There are no reported exploits in the wild at this time.

Patch status is not yet confirmed — check the vendor advisory at https://access.redhat.com/security/cve/CVE-2026-31230 for current remediation guidance. Until an official fix is available, avoid running the vulnerable script with untrusted input for the --clip_values and --input_shape arguments. Restrict access to pipeline configurations and automated scripts that invoke this component to trusted users only. Monitor the vendor advisory for updates on patches or official mitigations.