CVE-2026-31228 is a critical remote code execution vulnerability in the Adversarial Robustness Toolbox (ART) through version 1.20.1. The issue is in the Kubeflow component's robustness evaluation function for PyTorch models, which unsafely uses Python's eval() function to dynamically evaluate user-supplied strings for LossFn and Optimizer parameters without any sanitization or security restrictions. This allows an attacker to execute arbitrary Python code remotely, resulting in complete compromise of the system running the ART evaluation. The vulnerability is identified as CWE-94 (Improper Control of Generation of Code). The CVSS v3.1 base score is 9.8, reflecting its critical severity with network attack vector, no privileges required, no user interaction, and high impact on confidentiality, integrity, and availability. There is no explicit remediation level or patch information provided in the vendor advisory, and no known exploits in the wild have been reported as of the publication date.

Successful exploitation allows an unauthenticated remote attacker to execute arbitrary Python code on the system running the ART evaluation, leading to full system compromise. This includes complete loss of confidentiality, integrity, and availability of the affected system.

Patch status is not yet confirmed — check the vendor advisory at https://access.redhat.com/security/cve/CVE-2026-31228 for current remediation guidance. Until an official fix is available, avoid using the affected versions of ART in environments where untrusted input can reach the Kubeflow component's robustness evaluation function. Restrict access to the affected functionality and monitor for suspicious activity related to ART usage.