In this chapter, we’ll look at “robustness,” another important responsible AI principle and one of the principles in the SAFE-HAI framework. We will look at a brief definition of robustness and how it relates to the reliability of ML models, with specific reference to NIST’s risk management framework. We will also discuss the metrics for measuring robustness, with an introduction to the Rank Graduation “Robustness” score (RGR). We’ll conclude this chapter with a brief analysis of an adversarial benchmark for robustness, recommendations for mitigation, and a scoring rubric that could be adopted to analyze the robustness scores of your ML models/applications.