NIST AI RMF 1.0: organizing AI risk work (US · NIST AI RMF)

What category of voluntary framework this is

NIST AI 100-1 is the subject. The official title is Artificial Intelligence Risk Management Framework (AI RMF) 1.0. The NIST Artificial Intelligence Risk Management Framework (AI RMF) 1.0 is a voluntary framework. It is meant to preserve rights. It is non-sector-specific. It is use-case-agnostic. It helps organizations manage risks of AI systems. It also helps them incorporate trustworthiness considerations into design, development, use, and evaluation.

A common misread treats AI RMF certified as one seal. Another misread collapses AI RMF with the LLM Top 10 or the web Top 10. AI RMF 1.0 is a voluntary framework. No law requires it, and no scheme certifies against it.

Edition covered

The edition string is NIST AI 100-1, Artificial Intelligence Risk Management Framework (AI RMF) 1.0, published 26 January 2023 (2023-01-26). The DOI is 10.6028/NIST.AI.100-1. Official hubs are https://www.nist.gov/publications/artificial-intelligence-risk-management-framework-ai-rmf-10, https://www.nist.gov/itl/ai-risk-management-framework, and https://doi.org/10.6028/NIST.AI.100-1. On 2026-09-26, the publication page still stated Published January 26, 2023. It still listed report number NIST AI 100-1 and the same DOI. The NIST AI RMF home page still described voluntary use and the same DOI.

That home page states that AI RMF 1.0 is under revision as part of the White House AI Action Plan. That sentence is a process notice and not a published replacement edition. A future generation beyond 1.0 is UNKNOWN. If a newer Final is published, this pin must change. The older claim is then UNKNOWN.

Govern, Map, Measure, and Manage are literacy terms

Govern, Map, Measure, and Manage are the Core functions. Govern covers who is accountable. That scope includes policies, roles, and oversight culture for AI risk. Map covers what the system is, the context, and where harm can land. Measure covers how a team tests, evaluates, and tracks AI risks and trustworthiness. Manage covers what a team does with the result: prioritize, respond, control, and improve.

These four functions are jobs that organize AI risk work. They are not a finished test plan. Subcategory identifiers stay in the NIST publication. A sort of one control onto the four functions (Human oversight gates and the NIST AI Risk Management Framework) is practice, not a conformity assessment.

The function first, then the evidence

The habit starts with one real or synthetic AI-powered feature. One AI RMF function gives a shared term for the conversation. The conversation then moves to verifiable evidence. Examples are a specific human gate, tool authorization, a measure that can catch a miss, or a manage response when a measure fails.

NIST AI 600-1 is a companion, not this edition

NIST AI 600-1 is a later companion profile. Its official title is Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile. It was published July 26, 2024 (2024-07-26, DOI 10.6028/NIST.AI.600-1). It does not go into the AI RMF 1.0 edition string.

Separate NIST resources include the Playbook and the Roadmap. A Critical Infrastructure Profile exists only as a concept note (April 7, 2026).

Rewrite the one-line claim

The claim "we are AI RMF, LLM Top 10, Top 10, ASVS, and OWASP AI compliant" mixes five objects. Exact replacement: "We use NIST AI RMF 1.0 (AI 100-1, 2023-01-26) as a voluntary four-function organizer (Govern, Map, Measure, Manage)." That organizer is different from OWASP GenAI LLM Top 10 2026 awareness and from web Top 10:2025. The organizer also differs from ASVS 5.0.0 verification depth.

A slide can state "AI RMF certified" or "NIST AI verified." The team then checks three points: which edition applies, which function was evidenced on which system, and whether the live need is AI RMF placement, LLM awareness, web Top 10 awareness, or ASVS verification depth.

Four function names are not certified, not LLM Top 10, not web Top 10, and not ASVS

Eight phrases about AI RMF 1.0, each with its correction.
Phrase people use	Literacy correction
We listed Govern / Map / Measure / Manage, so we are AI RMF certified / complete	AI RMF 1.0 is a voluntary framework. A list of four functions is not a finished assessment and not a certificate.
AI RMF done means LLM Top 10:2026 done	OWASP GenAI LLM Top 10 2026 is a separate LLM-application awareness list. Explainer: /learn/explainers/owasp-llm-top-10/. The two objects stay separate.
AI RMF done means web Top 10:2025 done	OWASP Top 10:2025 is a separate web-application awareness list. Explainer: /learn/explainers/owasp-top-10/.
AI RMF done means ASVS Level N verified	ASVS is a separate requirements and verification project. Explainer: /learn/explainers/owasp-asvs/. Function names are not ASVS evidence.
We finished an AI security course / one function sort, so AI RMF is sealed	A course covers selected themes. One sort of one control does not finish the framework. Practice sort: /learn/topics/human-oversight-gates/#map-it.
We are AI RMF certified / NIST AI verified	AI RMF 1.0 is a voluntary framework, not a mail-order seal. Six-map comparison: /reference/frameworks/compare/.
We followed the Playbook, so AI RMF is complete	The Playbook offers suggested actions and is not a checklist to follow in full. The subcategory catalogs and Playbook actions stay in NIST resources. Field guide: /reference/frameworks/nist-ai-rmf-1-0/.
NIST AI 600-1 is the same edition pin as AI RMF 1.0	NIST AI 600-1 is a Generative AI Profile companion (July 26, 2024, DOI 10.6028/NIST.AI.600-1). It does not go into the 1.0 edition string.

Common misconceptions

Four function names mean AI RMF certified or an agent that is safe.

AI RMF 1.0 is a voluntary framework. A list of Govern, Map, Measure, and Manage is not a finished assessment and not a certificate.

AI RMF, the LLM Top 10, the web Top 10, and ASVS are the same thing.

They are different objects. NIST AI RMF 1.0 places work in Govern, Map, Measure, and Manage. OWASP GenAI LLM Top 10 2026 is a separate awareness list. Top 10:2025 covers web risk themes. ASVS is leveled verification depth. The four objects stay separate.

A finished AI security course, or one control sorted onto the four functions, means AI RMF is sealed.

One practice sort covers one control, not the whole framework.

A vendor homepage shield attests AI RMF completion.

A homepage shield is not evidence that a function was practiced on the system.

AI RMF 1.0 is a law that U.S. companies must follow.

AI RMF 1.0 is voluntary. No U.S. statute forces conformity to it.

NIST AI 600-1 is the AI RMF 1.0 edition pin.

NIST AI 600-1 is the Generative AI Profile, a companion published in 2024. The 1.0 pin stays NIST AI 100-1.

Govern, Map, Measure, Manage, as terms

These are four jobs that organize AI risk work. No row is a finished assessment.
Function	One-line literacy	Boundary
Govern	Who is accountable, with policies, roles, and oversight culture for AI risk	A job that organizes the work. Not a finished assessment.
Map	What the system is, the context, and where harm can land	A job that organizes the work. Not a finished assessment.
Measure	How a team tests, evaluates, and tracks AI risks and trustworthiness	A job that organizes the work. Not a finished assessment.
Manage	What a team does with the result. The actions are to prioritize, respond, control, and improve	A job that organizes the work. Not a finished assessment.

CHECK THE CATEGORY

Which sentence matches NIST AI RMF 1.0?

Glossary and nearby pages

Use the agency page in the sources for the authoritative text.