Loading
Released by the US National Institute of Standards and Technology on 26 January 2023 as AI RMF 1.0 (NIST AI 100-1), the framework is, in NIST's own words, voluntary, rights-preserving, non-sector-specific and use-case agnostic. It structures AI risk management into four functions, Govern, Map, Measure and Manage, and is referenced by regulators and enterprises far beyond the United States.
NIST AI Risk Management Framework, a voluntary, sector-agnostic framework published by the US National Institute of Standards and Technology that organises AI risk management around four functions: Govern, Map, Measure, and Manage.
The NIST AI RMF is the most widely adopted AI risk framework in the US enterprise market. Unlike the EU AI Act, it is not law: it is reference guidance. The Govern function is foundational and addresses organisational accountability for AI risk; the other three functions are operational. NIST has also published companion profiles for generative AI (NIST AI 600-1) and other high-risk contexts.
Source: NIST AI Risk Management Framework 1.0 (January 2023)
Establish AI risk management culture, policies, processes, and accountability structures across the organisation. GOVERN is cross-cutting and informs the other three functions.
Identify and categorise AI risks, context, stakeholders, potential harms, and applicable regulations.
Assess the magnitude of identified AI risks through quantitative and qualitative methods including bias testing.
Treat AI risks through controls, monitoring, incident response, and continuous improvement.
The framework anchors AI risk to seven characteristics of trustworthy AI (AI RMF 1.0, section 3). GOVERN, MAP, MEASURE and MANAGE exist to deliver systems that exhibit them:
Valid and reliable
The foundational characteristic: accuracy and consistent performance across the conditions the system will actually meet.
Safe
The system should not, under defined conditions, endanger human life, health, property or the environment.
Secure and resilient
Protected against adversarial examples, data poisoning, model exfiltration and other attacks; able to withstand and recover from disruption.
Accountable and transparent
Information about the system, its provenance, and responsibility for it is available to those who need it.
Explainable and interpretable
The mechanisms behind operation can be represented, and outputs carry meaning users can act on.
Privacy-enhanced
Guards against inference attacks that re-identify individuals or expose previously private information.
Fair, with harmful bias managed
Manages three bias categories, systemic, computational/statistical, and human-cognitive, which can cause discriminatory outcomes even without intent.
On 26 July 2024 NIST published a companion resource, the Generative AI Profile (NIST AI 600-1), produced under Executive Order 14110. It is a cross-sectoral profile of the AI RMF that sets out just over 200 suggested actions, mapped across the GOVERN, MAP, MEASURE and MANAGE functions, to help organisations manage the risks that are unique to or amplified by generative AI.
Those actions are organised against twelve categories of generative-AI risk. The profile is voluntary guidance rather than a checklist: organisations select the actions relevant to how they build or use generative AI, in the same risk-based spirit as the core framework.
Framework: NIST AI 100-1 (AI RMF 1.0) and NIST AI 600-1 (Generative AI Profile) · Last reviewed July 2026
This page is general information about the NIST AI RMF, not legal or compliance advice, and does not capture every nuance or update. Always verify against NIST's own published materials and your own qualified counsel before relying on it.