NIST AI 100-2e2025 Adversarial Machine Learning

Capture Summary

NIST Trustworthy and Responsible AI report defining adversarial machine learning taxonomy and terminology. The report organizes attack lifecycle stages, attacker goals, capabilities, knowledge, and mitigation concepts.

Relevance

• Foundational terminology source for Security for AI.
• Helps normalize evidence notes across evasion, poisoning, privacy, model extraction, and generative AI threats.
• Useful for mapping newer LLM/agent risks back to established AML concepts.

Collection Notes

Collected as official taxonomy source.