ISO/IEC 42001 for AI auditors and the AAIA

Of the standards an AI auditor meets, ISO/IEC 42001 is the one written as requirements: it says what an organization's management of AI must include, which makes it a natural yardstick for an audit. ISACA's AAIA exam content outline names no standard, but its topic on "Leading Practices, Ethics, Regulations, and Standards for AI" is where 42001 belongs, and six of our 18 AAIA lessons use it. This guide reads the standard as ISO describes it, from the auditor's side: what an AI management system is, how the standard works as criteria, how it differs from ISO's other AI standards, and what ISO/IEC 42006 adds for the audits that lead to certification.

The short answer

A requirements standard for managing AI. ISO says it "specifies requirements for establishing, implementing, maintaining, and continually improving an Artificial Intelligence Management System (AIMS) within organizations".
For anyone who builds, provides or uses AI. In ISO's words, "Organizations of any size involved in developing, providing, or using AI-based products or services."
A management system standard, run on Plan-Do-Check-Act. It covers the policies, objectives and processes that govern AI across an organization, not the details of one AI application.
The first edition is current. ISO/IEC 42001:2023 was published in December 2023, and ISO's page lists no amendment or newer edition.
Certification has a standard of its own. ISO/IEC 42006:2025 sets requirements for the bodies that audit and certify against 42001.
Not named by ISACA, but the kind of standard an auditor uses as criteria once an organization adopts it.

Why an AI auditor needs it

An audit finding is a gap between what is and what should be, and for AI the hard part is often the second half: what should an organization's management of AI look like? ISO/IEC 42001 answers with requirements, and ISO is explicit that this sets it apart from its other AI standards. Its page names three: "ISO/IEC 22989, which establishes terminology for AI and describes concepts in the field of AI; ISO/IEC 23053, which establishes an AI and machine learning (ML) framework for describing a generic AI system using ML technology; and ISO/IEC 23894, which provides guidance on AI-related risk management for organizations." And then: "ISO/IEC 42001 on the other hand is a management system standard (MSS)."

For an auditor, the difference is practical, and the reading here is ours. Terminology and guidance inform judgment; requirements can be tested. An organization that commits to 42001 has accepted a set of requirements an auditor can test it against, the way an organization that adopts the NIST AI RMF accepts a framework's outcomes. Our guides to the NIST AI RMF, the EU AI Act and the OWASP LLM Top 10 cover the other criteria an AI auditor meets, and ISACA's ITAF the standards for the auditor's own conduct.

What an AI management system is

ISO defines the thing the standard asks for. "An AI management system, as specified in ISO/IEC 42001, is a set of interrelated or interacting elements of an organization intended to establish policies and objectives, as well as processes to achieve those objectives, in relation to the responsible development, provision or use of AI systems."

Read as an auditor would, that definition names three kinds of evidence: policies, objectives, and the processes meant to achieve them. It also says they are "interrelated or interacting", which is the part an audit can test most usefully. An objective with no process behind it, a policy that no process carries out, or a process that measures nothing an objective cares about is a gap in the system, even when each document looks complete on its own. That reading is ours, but it follows from ISO's words.

A management system, not a model checklist

The standard works at the level of the organization, not the individual AI system. ISO says: "Implementing this standard means putting in place policies and procedures for the sound governance of an organization in relation to AI, using the Plan‐Do‐Check‐Act methodology. Rather than looking at the details of specific AI applications, it provides a practical way of managing AI-related risks and opportunities across an organization."

Two consequences follow for an audit, in our reading:

It tests the system, not each model. An audit against 42001 asks whether the organization governs its AI through a working management system. Whether a particular model is accurate, fair or secure needs its own criteria and its own tests, which our lessons on testing, supervision and threats cover.
It expects a cycle that turns. Plan-Do-Check-Act is a loop: plan, carry out, check the results, act on what the checks show. ISO's description adds "continually improving" to the standard's purpose. So an auditor looks for evidence that checks led to changes, not just that checks happened.

ISO also gives the standard's aim in its own terms: "it provides an integrated approach to managing AI projects, from risk assessment to effective treatment of these risks."

Who it applies to, and which AI

ISO's answer on scope is broad. The standard is for "Organizations of any size involved in developing, providing, or using AI-based products or services. It is applicable across all industries and relevant for public sector agencies as well as companies or non-profits." Asked whether it applies to all AI systems, ISO's page answers: "Yes, it's designed to be applicable across various AI applications and contexts."

That breadth matters to an auditor because it means 42001 fits organizations that only use AI they bought, not just those that build it. It also puts a planning question first: which of the organization's AI systems its management system covers. The standard's breadth does not settle that for any one organization, so an auditor confirms the stated coverage against the organization's own inventory of AI systems before relying on any claim of conformance.

Where 42001 sits among ISO's AI standards

ISO/IEC	What ISO says it does	What it gives an auditor (our reading)
22989	"establishes terminology for AI and describes concepts in the field of AI"	Shared terms for findings
23053	"establishes an AI and machine learning (ML) framework for describing a generic AI system using ML technology"	A way to describe the system under audit
23894	"provides guidance on AI-related risk management for organizations"	A reference for judging the risk process
42001	"specifies requirements" for an AI management system	Requirements to test conformance against
42006	Requirements for the bodies that audit and certify against 42001	A basis for weighing a certificate

ISO also sells 42001 in two packages: one with ISO/IEC 27001:2022, the information security management standard, and one with ISO/IEC 42005:2025. For an auditor, the first is a reminder that AI governance and information security can meet in the same organization, and that a security management system already in place may share processes with an AI one.

Certification and ISO/IEC 42006

Certification against 42001 is done by certification bodies, and since July 2025 ISO has a standard for them. ISO/IEC 42006 "sets out the additional requirements for bodies that audit and certify artificial intelligence management systems (AIMS) according to ISO/IEC 42001." It does not stand alone: "It builds on ISO/IEC 17021-1", and asked whether it replaces that standard, ISO answers: "No. It supplements ISO/IEC 17021-1 with AI-specific requirements for auditing and certification processes."

ISO names who should use it: "Certification bodies performing audits of AI management systems based on ISO/IEC 42001, as well as accreditation bodies assessing those certifiers." It also notes a wider use: "It can be integrated into broader conformity assessment schemes for AI products, processes or services under ISO/IEC 17065."

What this means for an AI auditor, in our reading:

An internal audit is not a certification audit. An organization's own auditors, or an external IT auditor engaged for assurance, evaluate against criteria for the organization's purposes. Certification is a separate service, done by a certification body under the rules of ISO/IEC 42006 and ISO/IEC 17021-1.
A certificate is evidence, not a conclusion. When an auditee presents a 42001 certificate, the auditor asks who issued it, what it covers, and when, and checks that its coverage includes the AI systems under audit. A certificate that covers one business unit says nothing about another.
Certification is not a precondition. An organization that has never been certified can still be audited against the standard's requirements, used as a benchmark.

The edition and the price

ISO's page gives ISO/IEC 42001:2023 as Edition 1, published in December 2023, 51 pages, at CHF 225, with no amendment or newer edition in its life cycle. ISO/IEC 42006:2025 is Edition 1 of July 2025, 31 pages, at CHF 159. Both come from the same committee, ISO/IEC JTC 1/SC 42. As with any criteria, name the edition in a finding: a gap against "ISO/IEC 42001:2023" stays clear if a later edition changes the requirements.

If you want the credentials built around the standard, such as a 42001 Lead Auditor certificate, our AIGP vs ISO 42001 and GARP RAI compares them with the AIGP.

How 42001 meets the AAIA's outline

ISACA's outline does not mention ISO/IEC 42001, and the mapping here is ours. Topic 1E, "Leading Practices, Ethics, Regulations, and Standards for AI", is its natural home. Among the tasks, evaluating "the organization's AI policies and procedures, including compliance with legal and regulatory requirements" and "whether the organization has defined ownership of AI-related risk, controls, procedures, decisions, and standards" are both questions a management system standard helps answer. Our lessons use 42001 in governance, risk management, regulation and standards, data, threats and audit planning, as a source of criteria rather than a subject to memorize.

How the standard shows up in a question

AAIA questions about standards rarely ask what a clause says. They ask how an auditor should use the standard:

How the exam asks it

An IS auditor is planning a review of a retailer's AI pricing tool. The retailer's own AI policy is thin, so the auditor will use ISO/IEC 42001 as the criteria. Which of the following is the BEST approach?

A. Select the standard's requirements that bear on the audit objective, and agree them in planning.
B. Test the retailer against every requirement in the standard, so that nothing is missed.
C. Postpone the audit until the retailer is certified, since only certification makes the standard usable.
D. Rely on the retailer's own policy alone, since a voluntary standard cannot serve as criteria.

Answer: A. Where an organization's own policy is thin, a recognized standard can supply the criteria, but the auditor applies the parts that bear on the engagement's objective rather than the whole standard indiscriminately, and settles them in planning, so the auditee knows the yardstick before fieldwork. Testing every requirement (B) spends the engagement on what its objective does not need. Certification (C) is not a precondition: an uncertified program can still be tested against the standard as a benchmark. And a voluntary standard (D) can serve as criteria once the auditor selects it for the engagement and agrees it in planning, which is exactly why it helps where policy is thin. Written for this guide in the exam's shape; it is not an ISACA question.

Where to start

Our AAIA lessons on governance, risk management and standards teach how 42001 and the other instruments supply audit criteria, and the lesson on audit planning teaches how to choose and scope them. A free account opens AI models and requirements in full, lessons, questions and flashcards; full access is a single payment with nothing to renew. For what the outline covers topic by topic, see the AAIA exam content outline explained.

By QualExam editorial. Written against the certifying body’s published documents, pinned by hash so a revised edition shows up as a change.