AI Governance Engineer
GitHub

NIST AI RMF vs ISO 42001

By Updated First published

Both are voluntary, and only ISO/IEC 42001 is certifiable. The NIST AI RMF 1.0 is a US-origin risk framework, non-sector-specific and use-case agnostic, organised in four functions: Govern, Map, Measure and Manage. ISO/IEC 42001 is the international AI management-system standard; certification bodies audit organisations against it, with their own competence set by ISO/IEC 42006.

At a glance

The two instruments side by side, as the Body of Knowledge states them. Each cell names the primary source it rests on; the last row links the chapter sections each line comes from.

Attribute NIST AI RMF ISO/IEC 42001
Type Framework: NIST AI Risk Management Framework 1.0 (NIST AI 100-1). Its companion Generative AI Profile (NIST AI 600-1, 2024) is not mapped on these pages. Sources: NIST AI 100-1, NIST AI 600-1 Standard: ISO/IEC 42001:2023, the AI management-system (AIMS) standard Source: ISO/IEC 42001:2023
Issuer NIST (United States) Source: NIST AI 100-1 ISO/IEC (JTC 1/SC 42) Source: ISO/IEC 42001:2023
Legal force Voluntary and US-origin. It describes itself as voluntary, rights-preserving, non-sector-specific and use-case agnostic. Source: NIST AI 100-1 Voluntary. It is a management-system standard, not the Article 17 QMS, and its European adoption confers no presumption of conformity with the AI Act. Sources: ISO/IEC 42001:2023, AI Act Art. 17, AI Act Art. 40
Scope and reach Any organisation, in any sector and for any use case. GOVERN applies across the whole process; MAP, MEASURE and MANAGE apply per system and per lifecycle stage. Source: NIST AI 100-1 Any organisation that develops, provides or uses AI. It specifies requirements for establishing, implementing, maintaining and continually improving an AI management system. Source: ISO/IEC 42001:2023
Certifiable No. There is no certification scheme for it: NIST AI 100-1 presents the RMF as voluntary guidance. Source: NIST AI 100-1 Yes. Certification bodies audit organisations against it; ISO/IEC 42006:2025 sets their additional requirements on top of ISO/IEC 17021-1. The certificate evidences a management system; it does not make a system compliant. Source: ISO/IEC 42006:2025
Key artefacts Four functions (Govern, Map, Measure, Manage) in 19 categories and their subcategories, used as control metadata; a current and a target profile, with the gap between them as the action plan. Source: NIST AI 100-1 Clauses 4 to 10 in the Harmonized Structure: AI policy, roles, AI risk assessment (6.1.2), risk treatment (6.1.3) and system impact assessment (6.1.4), internal audit, management review. Annex A control objectives in nine areas (A.2 to A.10), justified in a Statement of Applicability. Source: ISO/IEC 42001:2023
Dates 1.0 published 2023-01-26; there is no 2.0. A formal review was foreseen by 2028, and as of 2026-09-24 NIST states that 1.0 is being revised, with no revised version published. Sources: NIST AI 100-1, NIST: AI Risk Management Framework Published 2023. No application date: it applies to an organisation from the day it adopts the standard. Source: ISO/IEC 42001:2023
In the Body of Knowledge

Where they overlap, topic by topic

The crosswalk maps 25 AI governance topics. Both instruments file clauses under 20 of them, 11 strongly (a core clause on each side). 2 topics have a core clause only in the NIST AI RMF and 0 only in ISO 42001; 1 more is touched by one side only in passing (a related clause, not a core one); 2 are reached by neither. A shared topic means the two deal with the same thing, not that meeting one meets the other.

Strong: both file a core clause. Partial: both file a clause, at least one only in passing. Only, in passing: one side files a related clause and the other none. Clause ids link to their page in the obligation register where one exists. The last column names a pattern only where it serves a core clause on both sides: the crosswalk row's register entry lists it and the pattern's own "Maps to" line names that clause. Otherwise the cell is empty.

Topic What NIST AI RMF asks for What ISO 42001 asks for Overlap Patterns for both
Risk management
  • MAP 1 Context is established and understood
  • MAP 5 Impacts to individuals, groups, communities, organizations, and society are characterized
  • MANAGE 1 AI risks based on assessments and other analytical output are prioritized, responded to, and managed
  • 6 more in the crosswalk
Strong
Governance and accountability
  • GOVERN 1 Policies, processes, procedures, and practices across the organization related to the mapping, measuring, and managing of AI risks are in place, transparent, and implemented effectively
  • GOVERN 2 Accountability structures are in place so that the appropriate teams and individuals are empowered, responsible, and trained
  • GOVERN 4 Organizational teams are committed to a culture that considers and communicates AI risk
  • 1 more in the crosswalk
Strong
Impact assessment
  • MAP 3 AI capabilities, targeted usage, goals, and expected benefits and costs are understood
  • MAP 5 Impacts to individuals, groups, communities, organizations, and society are characterized
  • 6.1.4 AI system impact assessment
  • 8.4 AI system impact assessment (operation)
  • A.5 Assessing impacts of AI systems
Strong
Data governance
  • MAP 2 Categorization of the AI system is performed
  • MEASURE 2.10 Privacy risk of the AI system is examined and documented
  • MEASURE 2.11 Fairness and bias are evaluated and results are documented
  • A.7 Data for AI systems
  • A.7.3 Acquisition of data (not yet verified against the source)
  • A.4 Resources for AI systems
Partial
Documentation and transparency
  • MAP 1 Context is established and understood
  • MEASURE 2.8 Risks associated with transparency and accountability are examined and documented
  • MAP 1.6 System requirements are elicited from and understood by relevant AI actors. Design decisions take socio-technical implications into account to address AI risks
  • 1 more in the crosswalk
  • 7.5 Documented information
  • A.6 AI system life cycle
  • A.8 Information for interested parties
Partial
Inventory and registration
  • GOVERN 1.6 Mechanisms are in place to inventory AI systems and are resourced according to organizational risk priorities
  • GOVERN 1.7 Processes and procedures are in place for decommissioning and phasing out AI systems safely and in a manner that does not increase risks or decrease the organization's trustworthiness
  • A.4 Resources for AI systems
Strong
Logging and traceability
  • MANAGE 4 Risk treatments, including response and recovery, and communication plans for the identified and measured AI risks are documented and monitored
  • MEASURE 3 Mechanisms for tracking identified AI risks over time are in place
  • A.6 AI system life cycle
  • A.6.2.8 AI system recording of event logs (not yet verified against the source)
Partial
Human oversight
  • MANAGE 2.4 Mechanisms are in place and applied, and responsibilities are assigned and understood, to supersede, disengage, or deactivate AI systems that demonstrate performance or outcomes inconsistent with intended use
  • MAP 3.5 Processes for human oversight are defined, assessed, and documented in accordance with organizational policies from the GOVERN function
  • GOVERN 3.2 Policies and procedures are in place to define and differentiate roles and responsibilities for human-AI configurations and oversight of AI systems
Strong
Runtime guardrails
  • MANAGE 2 Strategies to maximize AI benefits and minimize negative impacts are planned, prepared, implemented, documented, and informed by relevant AI actors
Strong
Robustness, security and evaluations
  • A.6 AI system life cycle
  • 9.1 Monitoring, measurement, analysis and evaluation
Strong
Incident response and monitoring
  • MANAGE 4 Risk treatments, including response and recovery, and communication plans for the identified and measured AI risks are documented and monitored
  • MANAGE 4.3 Incidents and errors are communicated to relevant AI actors, including affected communities. Processes for tracking, responding to, and recovering from incidents and errors are followed and documented
  • MANAGE 2.4 Mechanisms are in place and applied, and responsibilities are assigned and understood, to supersede, disengage, or deactivate AI systems that demonstrate performance or outcomes inconsistent with intended use
  • 1 more in the crosswalk
  • A.8 Information for interested parties
  • 10.2 Nonconformity and corrective action
Strong
Supply chain and third parties
  • GOVERN 6 Policies and procedures are in place to address AI risks and benefits arising from third-party software and data and other supply chain issues
  • MAP 4 Risks and benefits are mapped for all AI system components including third-party software and data
  • MANAGE 3 AI risks and benefits from third-party entities are managed
  • 2 more in the crosswalk
  • A.10 Third-party and customer relationships
Strong
Prohibited practices
  • GOVERN 1.1 Legal and regulatory requirements involving AI are understood, managed, and documented
  • A.9.4 Intended use of the AI system (not yet verified against the source)
Partial
Fairness and non-discrimination
  • MEASURE 2.11 Fairness and bias as identified in the MAP function are evaluated and results are documented
  • GOVERN 3.1 Decision-making related to mapping, measuring, and managing AI risks throughout the lifecycle is informed by a diverse team
  • A.5.4 Assessing AI system impact on individuals or groups of individuals (not yet verified against the source)
Partial
Privacy and data protection
  • MEASURE 2.10 Privacy risk of the AI system as identified in the MAP function is examined and documented
  • A.7 Data for AI systems
Partial
Explainability and right to explanation
  • MEASURE 2.9 The AI model is explained, validated, and documented, and AI system output is interpreted within its context as identified in the MAP function to inform responsible use and governance
  • MEASURE 2.8 Risks associated with transparency and accountability as identified in the MAP function are examined and documented
  • A.8.2 System documentation and information for users (not yet verified against the source)
Partial
AI literacy and competence
  • GOVERN 2.2 The organization's personnel and partners receive AI risk management training to enable them to perform their duties and responsibilities consistent with related policies, procedures, and agreements
  • MAP 3.4 Processes for operator and practitioner proficiency with AI system performance and trustworthiness, and relevant technical standards and certifications, are defined, assessed, and documented
  • 7.2 Competence (not yet verified against the source)
  • 7.3 Awareness (not yet verified against the source)
Strong
Conformity assessment and certification
  • MEASURE 1.3 Internal experts who did not serve as front-line developers for the system and/or independent assessors are involved in regular assessments and updates
  • 9.2 Internal audit (not yet verified against the source)
Partial
IP and copyright
  • GOVERN 6.1 Policies and procedures are in place that address AI risks associated with third-party entities, including risks of infringement of a third-party's intellectual property or other rights
  • MAP 4.1 Approaches for mapping AI technology and legal risks of its components, including the use of third-party data or software, are in place, followed, and documented, as are risks of infringement of a third party's intellectual property or other rights
Not mapped NIST AI RMF only
Agent identity and autonomy
  • GOVERN 3.2 Policies and procedures are in place to define and differentiate roles and responsibilities for human-AI configurations and oversight of AI systems
Not mapped NIST AI RMF only, in passing
Sandboxes and real-world testing
  • MEASURE 2.3 AI system performance or assurance criteria are measured qualitatively or quantitatively and demonstrated for conditions similar to deployment setting(s)
  • A.6.2.4 AI system verification and validation (not yet verified against the source)
Partial
Environmental impact
  • MEASURE 2.12 Environmental impact and sustainability of AI model training and management activities as identified in the MAP function are assessed and documented
Not mapped NIST AI RMF only
Deployment, change and decommissioning
  • MANAGE 2.4 Mechanisms are in place and applied, and responsibilities are assigned and understood, to supersede, disengage, or deactivate AI systems that demonstrate performance or outcomes inconsistent with intended use
  • MANAGE 4.1 Post-deployment AI system monitoring plans are implemented, including mechanisms for capturing and evaluating input from users and other relevant AI actors, appeal and override, decommissioning, incident response, recovery, and change management
  • GOVERN 1.7 Processes and procedures are in place for decommissioning and phasing out AI systems safely and in a manner that does not increase risks or decrease the organization's trustworthiness
  • A.6.2.5 AI system deployment (not yet verified against the source)
  • A.6.2.6 AI system operation and monitoring (not yet verified against the source)
  • A.9 Use of AI systems
Strong

Can the NIST AI RMF help you certify to ISO/IEC 42001?

Partly. NIST itself hosts a crosswalk from the AI RMF to ISO/IEC 42001. But certification needs the management system itself, clauses 4 to 10 and a Statement of Applicability over the Annex A controls, which the RMF does not ask for. Its category and subcategory identifiers still serve as control metadata inside the AIMS.

Which should you start with?

Start with the NIST AI RMF to organise risk work: it is voluntary, non-sector-specific, and its four functions and 19 categories give you identifiers to tag controls with. Move to ISO/IEC 42001 when you need a certificate as proof of a working management system; the same identifiers then feed its Statement of Applicability.

Next step

Put the comparison to work on your own systems, in the browser.

Indicative, not legal advice and not a conformity claim. Nothing you enter leaves your browser.

Frequently asked questions

Is there an official crosswalk between the NIST AI RMF and ISO 42001?

NIST's AI Resource Center hosts crosswalks from the RMF to other frameworks, including ISO/IEC 42001, and dated 14 August 2025 a revised ISO/IEC 23894 crosswalk and a new ISO/IEC 42005 one. They are a sound starting point for a crosswalk file, not a substitute for mapping your own controls.

Source: NIST AIRC: crosswalks

Is there a NIST AI RMF 2.0?

No. AI RMF 1.0 (NIST AI 100-1, 26 January 2023) remains the citable text. As of 2026-09-24 NIST's framework page states that 1.0 is being revised as part of the White House AI Action Plan, but no revised version is published. Pin the version in control metadata.

Sources: NIST AI 100-1, NIST: AI Risk Management Framework

How do ISO/IEC 23894 and ISO 42001 relate to the NIST AI RMF?

ISO/IEC 23894 applies ISO 31000 risk management to AI; ISO/IEC 42001 is the certifiable management-system standard whose risk clauses (6.1.2 to 6.1.4, operated in 8.2 to 8.4) require the risk loop to exist and run. NIST's crosswalk shows that its functions and the 23894 clauses describe one process.

Sources: ISO/IEC 23894:2023, ISO/IEC 42001:2023, NIST AIRC: crosswalks

What does the NIST AI RMF cover that ISO 42001 does not?

In this crosswalk, 2 of the 25 topics have a core NIST AI RMF clause and no ISO 42001 clause mapped: IP and copyright; Environmental impact. Agent identity and autonomy is touched only in passing: the NIST AI RMF files a related clause there, not a core one, and ISO 42001 none. A topic with no ISO 42001 clause here is one this mapping does not reach, not one ISO 42001 is shown to leave out. The overlap table on this page lists the clauses; mappings are illustrative, not a claim of conformity.

Both instruments sit inside a wider field: AI governance, explained, from the laws and standards to the engineering practice.

Sources

Every clause on this page, with its note and verification status, is in the topic × framework crosswalk and its JSON download.

Other comparisons: ISO 42001 vs EU AI Act · NIST AI RMF vs EU AI Act