Skip to content

ISO/IEC 42001:2023 · AI management systems

ISO 42001: what it is, what it requires, and the documents you need

ISO/IEC 42001:2023 is the international standard for an AI management system - a framework for governing how an organization develops, provides or uses artificial intelligence responsibly. This is what it requires, the Annex A controls, a rollout checklist, and the exact document set you need.

Last updated

First document free. Card required; no charge today. No subscription.

Overview

What is ISO 42001?

ISO/IEC 42001:2023, published in December 2023, is the first international management-system standard for artificial intelligence. It defines the requirements for an AI management system (AIMS): the policies, processes and controls an organization uses to develop, provide or use AI responsibly, and to keep improving how it does so.

Like ISO 27001, it follows the Annex SL harmonised structure, so if you already run a management system the shape will be familiar - a policy, a risk-and-impact assessment, a set of controls chosen for your scope, and the records that prove the system is working.

Who needs ISO 42001?

Organizations that build, deploy, procure or embed AI systems - and increasingly those being asked by customers, procurement teams or regulators to show they govern AI responsibly. That includes a SaaS company shipping an LLM feature, an MSP managing AI tools for clients, or any business adopting AI in a decision that affects people.

Is ISO 42001 a certification?

The standard is the set of requirements; an accredited certification body can audit and certify an organization against it. Certification, its cost and any exam or training pathway are a separate job from producing the documentation - for that pathway, the Mindset Cyber family provides PECB-accredited ISO 42001 training and MindsetPrep covers exam preparation. PolicyMint produces the documentation you are audited against, not the certificate.

The requirements

The management-system clauses (4-10)

Clauses 4 to 10 are the mandatory core of the AIMS. They apply whatever your AI does. Here is what each asks, paraphrased, and the PolicyMint documents that satisfy it.

Clause What it asks Documents
Clause 4
Context of the organization
Define the scope of your AI management system and the internal and external issues, interested parties and requirements that shape it. AIMS scope & context
Clause 5
Leadership
Top management sets the AI policy, assigns roles and responsibilities and commits to the system. AI management policy · roles & responsibilities
Clause 6
Planning
Assess AI-related risks and opportunities, run AI impact assessments and set measurable objectives. AI risk assessment · AI impact assessment · objectives
Clause 7
Support
Provide the competence, awareness, communication and documented information the system needs. Competence & awareness · documented-information control
Clause 8
Operation
Put the planned controls and processes into practice across the AI system life cycle. AI system life-cycle procedures · operational controls
Clause 9
Performance evaluation
Monitor, measure, audit and review how well the system is working. Internal audit · management review · monitoring records
Clause 10
Improvement
Correct nonconformities and continually improve the system. Corrective action · continual-improvement records

Annex A

The ISO 42001 Annex A controls

Annex A lists 38 AI-specific controls, grouped into categories. You choose the ones that apply to your scope and record those decisions in the Statement of Applicability. The categories are:

Ref Category What it governs
A.2 Policies related to AI Setting and maintaining the organization's AI policy and its supporting policies.
A.3 Internal organization Accountability, roles and decision-making for AI systems.
A.4 Resources for AI systems The data, tooling, compute and people the AI depends on.
A.5 Assessing impacts of AI systems Assessing impacts on individuals, groups and society.
A.6 AI system life cycle Responsible design, development, deployment and retirement.
A.7 Data for AI systems How data is acquired, prepared and governed for AI.
A.8 Information for interested parties of AI systems What you tell users and affected parties about the AI.
A.9 Use of AI systems Intended, responsible use and human oversight.
A.10 Third-party and customer relationships Managing AI risk across suppliers, providers and customers.

See the full ISO 42001 Annex A controls list for every control by reference and title. PolicyMint decides which apply from your business profile and assembles your Statement of Applicability; to track implementation status control by control, our sister tool ControlStack maps the same controls in an interactive tracker.

The wedge

The ISO 42001 documents you actually need

The explainers rank; almost none of them tell you the actual document set. Here it is, tiered by what applies to you.

Required for every AIMS

  • AI management system policy
  • AIMS scope statement
  • AI risk assessment procedure & results
  • AI impact assessment procedure
  • Statement of Applicability (auto-assembled)
  • Roles, responsibilities & authorities
  • AI objectives & planning
  • Internal audit program
  • Management review records

Required if the control applies

  • Data management for AI policy
  • AI system life-cycle / development procedure
  • Human oversight procedure
  • Supplier & third-party AI policy
  • Transparency & information-to-users policy
  • AI incident management procedure
  • Competence & training records

Supporting / optional

  • Responsible-AI / ethics statement
  • Model & dataset documentation templates
  • Change management procedure
ISO 27001 documents

36

ISO 27001 documents

tiered by what you actually need

ISO 42001 documents

21

ISO 42001 documents

for AI management systems

Clauses and controls mapped

182

Clauses and controls mapped

every document cites the ones it satisfies

Verifiers per section

3

Verifiers per section

on a different model to the writer

The checklist

ISO 42001 checklist: the seven steps and what each one must produce

  1. 1

    Define scope and inventory your AI

    Decide which AI systems the management system covers and build an inventory of them.

    • A scope statement naming the AI systems, uses and business units covered
    • An AI system inventory: each system, its purpose, its owner and the data it touches
  2. 2

    Secure leadership and set the AI policy

    Get top-management commitment and publish the AI policy that everything else hangs from.

    • A signed AI policy, with the roles and responsibilities it assigns
    • A record of management commitment: objectives, resources and who is accountable
  3. 3

    Run AI risk and impact assessments

    Assess risks to the organization and impacts on people, and decide how you will treat them.

    • An AI risk assessment with a stated method, criteria and a risk register
    • An AI system impact assessment for each system in scope, covering individuals, groups and society
    • A risk treatment plan with owners and dates
  4. 4

    Select Annex A controls and build the SoA

    Choose the controls that apply to your scope and record the decisions in the Statement of Applicability.

    • A Statement of Applicability listing every Annex A control with an included-or-excluded decision and a justification
  5. 5

    Produce the document set

    Turn those decisions into the policies, procedures and records an auditor will assess.

    • The policies, procedures and records the SoA points to, each cited to the clause or control it satisfies
    • A document register with owners, versions and review dates
  6. 6

    Operate, record and train

    Run the controls day to day, keep the evidence, and make sure people know their part.

    • Operating records: monitoring logs, incident and change records, supplier reviews
    • Training and awareness records for the people the policy names
  7. 7

    Audit, review and (optionally) certify

    Run an internal audit and management review; an accredited certification body can then audit you against the standard.

    • An internal audit report with findings and corrective actions
    • Management review minutes covering performance, risks and improvements

ISO 42001 audit checklist: what the auditor asks to see

The outputs above are what an audit is made of. In roughly the order they come up:

  1. 01 The scope statement and the AI system inventory it applies to
  2. 02 The signed AI policy and the roles it assigns
  3. 03 The AI risk assessment method and the current risk register
  4. 04 An AI system impact assessment for each system in scope
  5. 05 The Statement of Applicability, with a justification for every exclusion
  6. 06 The documents each included control points to, retrievable by name
  7. 07 Records that the system runs: monitoring, incidents, supplier reviews, training
  8. 08 The internal audit report and management review minutes

Two standards

ISO 42001 vs ISO 27001

ISO 27001 governs information security (an ISMS). ISO 42001 governs AI management (an AIMS). They share the same Annex SL structure, so the management clauses (4-10) line up and the two integrate cleanly - many organizations run both, with shared clauses and separate Annex A control sets. If AI is part of how you handle information, they reinforce each other rather than compete.

PolicyMint generates the document set for both ISO 27001 and ISO 42001, and keeps overlapping documents consistent across the two systems so you are not maintaining the same policy twice. For a fuller breakdown, see ISO 42001 vs ISO 27001. Just need the AI rules for staff first? Our AI policy template guide walks that document section by section, with what an auditor checks and the clauses teams get wrong. The two assessments the standard asks for each have a free template too: the AI risk assessment template and the AI impact assessment template. In Australia, the free AI6 to ISO 42001 crosswalk maps the National AI Centre's Guidance for AI Adoption onto the same documents.

How it works

How PolicyMint generates your ISO 42001 documents

A template pack hands you blank files. Prompting a chatbot invents controls you never bought and is equally confident whether it is right or wrong. PolicyMint is different: it is grounded in the licensed clause text for the exact clauses each document maps to, it knows the full ISO 42001 document set and tells you which ones you actually need, and it checks its own work with three independent verifiers running on a different AI model before you ever see the draft - then exports everything in your branding.

PolicyMint's Full ISMS / AIMS pack is a one-time US$229 - A$350 including GST - and covers a complete certifiable ISO 42001 set of 21 finished documents, or a complete ISO 27001 set of 36. Each document is drafted from your own business profile, checked by three independent verifiers before you see it, and exported in your branding as Word, PDF and Excel. There is no subscription and credits are valid for 18 months.

See the full comparison →

ISO 42001 FAQ

ISO 42001, answered

The questions people ask before they start.

What is ISO 42001 in simple terms?

ISO/IEC 42001:2023 is the international standard for an AI management system - a framework for governing how an organization develops, provides or uses artificial intelligence responsibly. It sits alongside standards like ISO 27001 and uses the same management-system structure.

When was ISO 42001 published?

ISO/IEC 42001 was published in December 2023. It is the first international management-system standard specifically for artificial intelligence.

Is ISO 42001 a certification?

The standard is a set of requirements; an accredited certification body can audit and certify an organization against it. Certification, its cost and any exam or training pathway are a separate job from producing the documentation - for that pathway see the Mindset Cyber family (PECB-accredited training) and MindsetPrep. PolicyMint produces the documentation you are audited against, not the certificate.

How is ISO 42001 different from ISO 27001?

ISO 27001 governs information security (an ISMS); ISO 42001 governs AI management (an AIMS). Both share the same Annex SL structure, so they integrate cleanly - many organizations run both, with shared management clauses and separate Annex A control sets.

What documents do you need for ISO 42001?

At minimum: an AI management policy, defined scope, an AI risk assessment, an AI impact assessment, a Statement of Applicability and the records that show the system is running. Further Annex A policies apply depending on your scope. PolicyMint works out which of its ISO 42001 documents you actually need and drafts them for your business.

Do I need ISO 27001 before ISO 42001?

No. You can pursue ISO 42001 on its own. Because the two standards share a management-system structure, an existing ISO 27001 program gives you a head start, but it is not a prerequisite.

Does PolicyMint get me ISO 42001 certified?

It produces the documentation, not the certificate. A certification body audits your organization, not your PDFs. PolicyMint gives you a complete, consistent, credible ISO 42001 document set, grounded in the standard and written for how you actually operate.

Ready when you are

See what “finished” feels like.

Set up your business once, create your first ISO 42001 document free, and review the real output before you buy a pack.

Card required to download; no charge today. No subscription to cancel, no sales call.