The world's first certifiable AI management system standard, mapped for the classroom.
ISO/IEC 42001 tells an organization how to run the ongoing business of managing AI responsibly — policies, roles, risk processes, and controls — structured the same way as ISO 27001 and ISO 9001, so it slots into an organization's existing management systems rather than replacing them.
The AIMS runs on Plan–Do–Check–Act. Leadership (Clause 5) sits at the hub — every quadrant answers to it, and the wheel turns continuously, not once.
Purpose & Origin
What the standard is, why it exists, and what "certifiable" actually means here.
What it is
ISO/IEC 42001:2023 is the first international standard specifying requirements for establishing, implementing, maintaining, and continually improving an AI Management System (AIMS) within an organization. It was published in December 2023 by ISO and IEC's joint technical committee (JTC 1/SC 42), the same body that works on AI standards broadly.
Why it was created
Organizations building or using AI needed a management-system standard that plugs into what they already run — quality management (ISO 9001), information security (ISO/IEC 27001) — rather than a bespoke AI-only process. 42001 gives auditors, regulators, and business partners a common, checkable structure for "does this organization actually manage its AI risk," not just "does its AI perform well."
Who it's for
Any organization that provides or uses products or services utilizing AI systems, regardless of size or sector — from a startup shipping one model to a multinational with dozens of AI-enabled products across its supply chain.
What "certifiable" means
Unlike the NIST AI RMF, which is voluntary guidance, an organization can hire an accredited third-party certification body to audit its AIMS against 42001 and issue a certificate — similar to how organizations get ISO 27001-certified for information security today.
How to Use This Guide
This tool summarizes a management-system standard for a course session — it is not a substitute for an implementation project or a certification audit.
- Start with the PDCA wheel. Everything below sits somewhere on Plan–Do–Check–Act. If you remember which quadrant a clause lives in, you'll remember what it's for.
- Read Clauses 4–10 as one continuous document. Unlike NIST AI RMF's iterative functions, these clauses are a genuine numbered sequence — each builds on commitments made in the one before it (you can't do Clause 8 Operation without the scope Clause 4 defined).
- Treat Annex A as "what," not "how." Annex A controls are reference points an auditor checks for evidence against — they are deliberately terse. The organization decides how to actually implement each one.
- Compare it against NIST AI RMF. Use the comparison table below — many programs run both: NIST AI RMF for the risk-thinking, ISO/IEC 42001 for the certifiable management structure around it.
- Cite the primary standard for graded work. ISO standards are copyrighted and must be purchased from ISO or a national standards body — this guide paraphrases publicly available structure and commentary, not licensed clause text.
ISO/IEC 42001 vs. NIST AI RMF
Different tools for a related job — worth knowing which one a given task actually calls for.
| ISO/IEC 42001:2023 | NIST AI RMF 1.0 | |
|---|---|---|
| Type | Certifiable management system standard | Voluntary risk-management framework |
| Structure | Numbered clauses 4–10 + Annex A controls | Four functions: Govern, Map, Measure, Manage |
| Origin | ISO/IEC JTC 1/SC 42 (international) | U.S. NIST (national, but widely adopted) |
| Primary question | "Can we prove, to an auditor, that we manage AI risk?" | "How do we think about and reduce AI risk?" |
| Typical pairing | Sits alongside ISO 27001, ISO 9001 | Sits alongside NIST CSF, Privacy Framework |
Clause 4 — Context of the Organization
Before anything else, the organization has to define what the AIMS actually covers — its boundaries, its stakeholders, and the issues that could affect it.
Internal and external issues relevant to the organization's purpose that affect its ability to achieve the AIMS's intended outcomes — market pressures, technical capability, regulatory environment.
Identifying who has a stake in the organization's AI — customers, regulators, employees, affected communities — and what each expects.
A documented, bounded statement of which AI systems, business units, and activities the management system actually covers — the single most consequential decision in an implementation, since everything else is scoped to it.
The organization establishes, implements, maintains, and continually improves the AIMS as an actual functioning system — not a document that sits on a shelf.
Clause 5 — Leadership
Top management has to own the AIMS visibly — a management system with no leadership commitment is the most common reason implementations fail an audit.
Top management demonstrates leadership by integrating AIMS requirements into business processes, ensuring resources are available, and communicating the importance of effective AI management.
A documented, top-management-approved policy stating the organization's commitments around responsible AI — communicated internally and available to interested parties.
Specific roles for the AIMS are assigned and communicated — someone is accountable for conformity, someone reports on performance to top management.
Clause 6 — Planning
Where AI-specific risk assessment and impact assessment enter the standard — this is the clause that most distinguishes 42001 from a generic management-system standard.
Includes a formal AI risk assessment process and an AI system impact assessment process — evaluating potential consequences for individuals and society, not only the organization.
Measurable objectives, consistent with the AI policy, with a plan for who does what, by when, with what resources.
Changes to the AIMS are made in a planned manner — an organization doesn't get to quietly redefine its scope or process mid-cycle.
Clause 7 — Support
The resourcing clause — people, competence, awareness, communication, and documentation that make the plan executable.
The organization determines and provides the resources needed for the AIMS, and ensures people doing AI-related work under its control are competent — through education, training, or experience.
Relevant people are aware of the AI policy and their contribution to the AIMS's effectiveness; internal and external communications about the AIMS are planned, not ad hoc.
Required records — policies, risk assessments, impact assessments, audit results — are created, controlled, and kept available; this is what an auditor actually reviews.
Clause 8 — Operation
Where the plans from Clause 6 actually get carried out, across the AI system lifecycle.
Processes needed to meet AIMS requirements are planned, implemented, and controlled — including managing planned changes and reviewing unintended ones.
Performed at planned intervals or upon significant change — not just once before launch.
Implementing the risk treatment plan developed in Clause 6 — the point where risk decisions become real action.
Carried out for individual AI systems as they're developed or before significant changes — evaluating impact on individuals, groups, and society.
Clause 9 — Performance Evaluation
The clause an external auditor will spend the most time on — proof that the organization actually checks its own work.
The organization determines what needs monitoring, the methods used, and when results are analyzed — evidence-based, not anecdotal.
Planned internal audits, conducted by people independent of what they're auditing, to check the AIMS conforms to the organization's own requirements and the standard's.
Top management reviews the AIMS at planned intervals to ensure it remains suitable, adequate, and effective — closing the loop back to Clause 5 Leadership.
Clause 10 — Improvement
What happens when Clause 9 finds something wrong — and how the whole cycle feeds back into Clause 4.
The organization continually improves the AIMS's suitability, adequacy, and effectiveness — the standard expects the system to keep getting better, not just stay compliant.
When something doesn't conform, the organization reacts to it, evaluates the need for action to eliminate the cause, and reviews whether the corrective action actually worked.
Annex A — Controls
Annex A works like ISO/IEC 27001's Annex A: a reference set of controls an organization selects from, based on its own risk assessment, and documents in a Statement of Applicability. These are terse by design — Annex A says what, not how.
Policies related to AI
AI policy content, review, and alignment with other organizational policies.
Internal organization
Roles and responsibilities specific to AI, reporting lines, segregation of duties.
Resources for AI systems
Data, tooling, system, and human resources allocated to AI systems and their documentation.
Assessing impacts of AI systems
Impact assessment process, considering effects on individuals, groups, and society, including environmental impact.
AI system life cycle
Controls spanning design, development, verification and validation, deployment, operation, monitoring, and decommissioning.
Data for AI systems
Data quality, provenance, and acquisition practices feeding AI systems.
Information for interested parties
Transparency obligations — what's communicated to users, regulators, and the public about a given AI system.
Use of AI systems
Responsible use practices, including use of third-party AI systems within the organization.
Third-party and customer relationships
Controls covering suppliers, partners, and customer-facing commitments around AI.
Classroom Uses
Ways instructors have used this structure with students.
Gap-analysis exercise
Give students a fictional (or their own institution's) AI use case and have them identify, clause by clause, what documentation would already exist versus what would need to be created from scratch.
Statement of Applicability drill
Assign a handful of Annex A controls and have students argue whether each is "applicable" to a given scenario and why — mirroring the real Statement of Applicability exercise auditors expect.
Mock internal audit
Pair students up: one drafts evidence for a clause (e.g., Clause 9 monitoring records), the other plays internal auditor and probes for gaps.
Cross-framework mapping
Have students map one 42001 clause or Annex A control to the closest NIST AI RMF category, then discuss where the mapping is clean and where it isn't.
Sources & Further Reading
This guide is a teaching summary. ISO standards are copyrighted texts — cite and quote from the official standard for anything graded.
ISO/IEC 42001:2023
The official standard, purchasable through ISO or a national standards body.iso.org — search "ISO/IEC 42001"
ISO/IEC JTC 1/SC 42
The joint technical committee developing 42001 and related AI standards — useful for context on where 42001 sits among other AI standards work.iso.org/committee/6794475.html
Accredited certification bodies
For understanding what an actual certification audit involves, check national accreditation body listings for organizations authorized to certify against 42001.