On 9 November 2026, the US National Institute of Standards and Technology (NIST) released an initial public draft of NIST SP 2400-102e2026, Adversarial Machine Learning: A Taxonomy and Terminology of Attacks and Mitigations. Public comments are open until 18 December 2026.

NIST says the report provides a taxonomy of concepts and defines terminology in adversarial machine learning, and that this edition focuses on attacks on agentic systems and the corresponding mitigations. The taxonomy is arranged by key types of machine-learning methods, stages of the attack life cycle, and attackers’ goals, objectives, capabilities and knowledge. A new agentic attacks taxonomy index allows fine-grained referencing of specific threats.

According to NIST, the draft draws on an extensive literature review, conversations with experts and research by NIST’s own specialists. It is intended to establish a common language that can inform other standards and future practice guides for assessing and managing the security of these systems. The intended audience includes those who design, develop, deploy, evaluate and govern them. The draft is not a binding regulation, and comments can be sent to the address given in the document.

Note from Evangelii Network Japan: this paragraph is our own comment, not part of NIST’s announcement. Clear shared terms for AI risks support the kind of human oversight and accountability discussed in our paper on Catholic AI and human dignity.

Sources