Artificial Intelligence (AI) is showing enormous promise in research and low-risk commercial settings. However, current AI models are often brittle, opaque, and fail to live up to performance expectations once deployed in the real world, where conditions are dynamic and uncertain. Achieving Trustworthy AI—AI that is safe & secure, accountable, fair & privacy-preserving, explainable and robust/reliable—is critical to fully leveraging AI in settings with strict reliability, safety, legal, and ethical requirements.