The statement accurately describes a key practice in MLOps and AI security. Integrating automated adversarial testing into a CI/CD (Continuous Integration/Continuous Deployment) pipeline enables proactive security validation. By programmatically generating and testing the model against perturbed inputs (e.g., via evasion attacks), teams can continuously measure and enforce a baseline for adversarial robustness before a new model version is deployed. This shifts security testing "left," making it an integral part of the development lifecycle.

This is a True/False question, so there are no other options to evaluate.

1. MITRE. (2023). MITRE ATLAS (Adversarial Threat Landscape for Artificial-Intelligence Systems). Technique: T0029 - Evasion Attack. The ATLAS framework documents evasion attacks and recommends defensive measures, including robustness testing, which aligns with the practice described.

2. O'Reilly Media. (2020). Building Secure and Reliable Systems. Chapter 21, "Security in the Development Lifecycle," discusses the importance of integrating security testing into automated build and deployment pipelines (CI/CD). This principle extends directly to AI/ML systems.

3. Bhatt, U., et al. (2020). Uncertainty as a Form of Transparency: Measuring, Communicating, and Using Uncertainty in Machine Learning. Proceedings of the 2020 AIES Conference. This and similar academic works highlight the need for rigorous model evaluation, including robustness to distributional shifts and adversarial examples, which is facilitated by automated testing. https://doi.org/10.1145/3375627.3375814