A set of properties, defined by the NIST AI RMF, is used to characterize the level of confidence in an AI system's performance and behavior. These characteristics form the basis for measuring and managing AI risks throughout the system's lifecycle.

‍