R&D support for secure AI at NIST's NCCoE — a federal contract (USAspending)

The NCCoE (National Cybersecurity Center of Excellence) is a public-private collaboration hub operated by NIST (the U.S. National Institute of Standards and Technology). Its role is to demonstrate practical cybersecurity through reference implementations and practice guides that organizations can actually use. "Secure AI" here refers to the question of how to protect AI models and the systems built around them from threats such as malfunction, tampering, data leakage, and adversarial inputs. As AI spreads into government, healthcare, and finance, there is growing need to treat the underlying models and data themselves as potential attack surfaces.

What makes this work notable is that it addresses AI not in terms of "capability" but in terms of "robustness and defensibility," approached through the lens of standardization. NIST is not a regulator but a standards and guidelines body, so its output tends to propagate not as a single product but as a shared yardstick the whole industry can reference. If evaluation methods and design best practices for secure AI are consolidated, they can influence not only federal procurement but also how private-sector systems are designed.

The specific research themes, target models, and deliverables of this contract are not stated in the source description, so we do not go further. Viewed across the field, it illustrates how, with the spread of generative AI and machine learning, the focus of security is expanding from "networks and endpoints" toward "the lifecycle of models and data."