Safety and alignment

NIST publishes adversarial machine-learning risk taxonomy

NIST published AI 100-2e2025, a standardized taxonomy covering evasion, poisoning, privacy and generative-AI misuse attacks plus mitigations and their limitations.

CURRENT ASSESSMENT · REVISION 1
AWAY FROM DOOM29confidence 97/100

Why it moved the index

The official taxonomy gives developers and governors a shared operational framework for identifying and mitigating attacks across AI systems. Impact is bounded because the guidance is voluntary.

0 comments · 0 votesOpen discussion

Public discussion is readable by everyone. Sign in to comment, reply, or vote.

No comments yet. Start the discussion.

AUDIT TRAIL

Assessment history

  1. R1
    Away 29 · confidence 97

    New March 2025 national AI-security guidance with no durable collision.

    12 Aug 2026