Compliance · ISO/IEC 42001 ISO/IEC 42001:2023 · Annex A — 38 controls, 9 objectives

ISO/IEC 42001 AI Management System Readiness

The world's first certifiable standard for governing artificial intelligence. We help you build an AI management system that holds up — designing it with your teams, testing the AI systems it governs the way an adversary would, and bringing you to audit-ready for a clean Stage 1 and Stage 2 certification audit.

Gap assessment to internal audit. CERT-In Empaneled. The accredited body issues the certificate — we make sure you earn it.

ISO/IEC 42001:2023 Annex A — 38 controls, 9 objectives Clauses 4–10 Integrates with ISO 27001 CERT-In Empaneled CISA-certified IS auditors in-house
Standard
ISO/IEC 42001:2023
Controls
38 Annex A controls · 9 objectives
Role
Certification-ready — the accredited body issues the certificate
Validation
Adversarial testing of the AI systems the AIMS governs
02

The first management standard built for AI

ISO/IEC 42001:2023 is the first international standard for an Artificial Intelligence Management System — an AIMS — and the first that an organization can be independently certified against.

It is not a checklist of model-safety techniques; it is a framework for governing AI as an ongoing organizational discipline. You define the scope of the AI systems you build, provide or use, assess the risks they create, assess their impact on the people and societies they touch, decide which controls treat those risks, operate them, and continually improve — under the oversight of leadership and the scrutiny of internal and external audit. Certification is independent, third-party proof that this system exists and works.

Published in December 2023 by ISO/IEC JTC 1/SC 42, the standard arrived as AI moved from experiment to infrastructure — and as regulators, boards and enterprise buyers started asking a harder question than "does it work?" They now ask "can you show it is governed responsibly?" ISO 42001 is built to answer that. It follows the same harmonized high-level structure as ISO/IEC 27001, so its management-system clauses, terms and core requirements line up — which means it integrates cleanly with an existing ISMS rather than duplicating it. Two requirements set it apart from a classic management standard: it mandates both an AI risk assessment and an AI system impact assessment — the second a deliberate reckoning with the consequences of your AI for individuals, groups and society, not just for the organization.

"Anyone can publish an AI policy. ISO 42001 asks you to prove the system behind it works — so we build the system to be true, and then we test the AI it governs."
03

Why organizations pursue ISO 42001

ISO 42001 applies to any organization that develops, provides or uses AI — regardless of size or sector. It is voluntary, but the pressure to adopt it is not.

Criterion Why it matters
You build or sell AI products Enterprise buyers and procurement teams increasingly ask AI vendors to demonstrate governed, responsible development before they will sign — and a certificate answers the question once.
You use AI in decisions that affect people Where AI informs hiring, lending, claims, pricing or safety, a certified AIMS shows regulators and the public that the impact was assessed, not assumed.
Regulation is arriving The EU AI Act and a wave of national AI rules reward demonstrable governance. ISO 42001 gives you a recognized, auditable system to point to.
Your board wants assurance on AI Independent certification gives leadership and audit committees objective evidence that AI risk is governed, not left to individual teams.
You already run ISO 27001 Because 42001 shares the same high-level structure, you can extend an existing ISMS to AI governance — assess once, reuse the management system, certify both.
You want to lead on trust As the first certifiable AI management standard, ISO 42001 is a credible, early signal that your organization takes responsible AI seriously.
04

What the standard actually asks for

ISO/IEC 42001:2023 has two halves. The management-system clauses (4–10) define how the AIMS is run — context, leadership, planning, support, operation, performance evaluation and improvement — including the standard's two signature requirements, an AI risk assessment and an AI system impact assessment.

Annex A then provides the reference catalogue of AI management controls: 38 controls organized under 9 control objectives (A.2 through A.10), with implementation guidance for every one in the normative Annex B. Your Statement of Applicability records, control by control, which you apply and why — the single most scrutinized document in the audit.

05

From gap to certification-ready — and beyond

Certification is awarded by an accredited certification body after a two-stage audit. Our job is everything that earns it: building an AIMS that is genuinely sound, assessing the risk and impact of the AI it governs, validating that the controls work, and assembling the evidence so the audit confirms what is already true.

We take you from a first gap assessment to a confident Stage 2 — and stay with you through surveillance and recertification.

01 · Intect
Gap Assessment
We measure your current AI governance against ISO/IEC 42001:2023, clause and control, and produce a clear gap report with a prioritized roadmap to close it. Where you already hold ISO 27001, we map what carries over.
02 · Intect
AI Risk + AI Impact Assessment
We run the standard's two signature exercises: an AI risk assessment and risk treatment plan, and an AI system impact assessment that documents the consequences of your AI for individuals, groups and society.
03 · Intect
AIMS Implementation
We build the AI policy and procedure suite, stand up the Annex A controls across the AI life cycle, data, use and third parties, and support remediation of every gap from step one.
04 · Intect
Statement of Applicability
We produce the SoA — each Annex A control included or excluded with documented justification — the document the auditor returns to most.
05 · Intect
Internal Audit
We conduct the internal audit the standard requires, independently checking that controls are not just documented but operating. Our in-house CISA-certified IS auditors give this independent rigour.
06 · Intect
Pre-Assessment (Mock Audit)
We run a mock Stage 1 and Stage 2 — auditing you the way the certification body will, so there are no surprises on the day.
07 · Accredited body
Stage 1 Audit (Documentation)
Conducted by the accredited certification body. A readiness and documentation review of your AIMS, scope and SoA. We support you through it.
08 · Accredited body
Stage 2 Audit (Implementation)
Conducted by the accredited certification body. The in-depth assessment that tests whether your controls genuinely operate. On success, the certificate is issued.
09 · Ongoing
Surveillance & Recertification
Certification runs on a three-year cycle: annual surveillance audits keep it live, and a full recertification audit renews it. We support continuous compliance across the whole cycle — essential where your AI systems keep learning and changing.

FIG. 02 — AIMS & certification lifecycle · 01–06 Intect-led · 07–08 accredited certification body · 09 the three-year cycle

The whole cycle is a Plan-Do-Check-Act loop — risk and impact assessment and implementation (Plan/Do), internal audit and management review (Check), remediation and continual improvement (Act). The AIMS is built to get better with every pass — which matters more for AI than for almost anything else, because the systems themselves don't stand still.

06

What an engagement includes

We scope to where you are — a first-time AIMS, an extension of an existing ISO 27001 ISMS to AI governance, or sustaining an existing certificate — and cover the work end to end.

Assess
01

Gap assessment against ISO/IEC 42001:2023

(management clauses 4–10 + all 38 Annex A controls)

02

AIMS scope definition

which AI systems, products, services, sites and third parties are in scope

03

AI risk assessment and risk treatment plan

04

AI system impact assessment

consequences for individuals, groups and society

Build
05

Statement of Applicability

inclusion/exclusion decisions with justification, against Annex A

06

AI policy and procedure suite aligned to the standard

07

Control-implementation support across the AI life cycle, data, use and third-party relationships

08

Remediation of identified gaps, with a prioritized roadmap

09

Integration with an existing ISO 27001 ISMS where one exists, to avoid duplicated effort

Validate (the offensive edge)
10

Adversarial testing of the governed AI systems

prompt injection, model abuse, jailbreaks, data-poisoning exposure and related attacks — that proves controls actually work

11

Technical control validation and configuration review of the AI/ML stack and supporting infrastructure

12

Evidence collection and an audit-ready artifact pack

13

Remediation retest after fixes

Prepare for audit
14

Internal audit, conducted to the standard's requirements

by CISA-certified IS auditors

15

Management-review pack for leadership oversight

16

Pre-assessment / mock Stage 1 and Stage 2

17

Liaison and support through the certification body's Stage 1 and Stage 2 audits

Sustain
18

Surveillance-audit support across the three-year cycle

19

Recertification readiness

20

Continuous-compliance guidance as your AI systems, risks and controls evolve

07

We govern AI — then we attack it

A paper audit confirms an AI control is written down. Because Intect is an offensive-security firm, we confirm the AI system it governs actually withstands attack.

Controls that are written

Most AIMS work governs AI on paper

The conventional path to ISO 42001 documents policies, samples records and ticks the controls that have evidence behind them. That satisfies the letter of the standard — but a documented control for "responsible use" and a model that an attacker can jailbreak into unsafe behaviour are not the same thing. The risks that never surface in a document review — prompt injection, training-data poisoning, model extraction, unsafe tool use by an autonomous agent — are exactly the risks ISO 42001 exists to manage.

Controls that are proven

We test the AI the management system governs

Intect is CERT-In Empaneled with an offensive-security heritage, so our readiness work is technically validated against the actual AI systems in scope. Our researchers probe them safely, the way an adversary would — prompt injection and jailbreaks, model abuse and extraction, data-poisoning exposure, and the new attack surface of autonomous agents — and bring you pentest-backed evidence that your Annex A controls hold. This is the same capability behind our AI/ML Penetration Testing and AI-Assisted & Autonomous Penetration Testing services. You walk into Stage 2 with an AIMS whose controls have already withstood a real probe, and you leave governing AI that means it.

The certificate proves you govern AI. Our testing proves the AI you govern can take a hit.

AI ADVERSARIAL TESTING — PENTEST-BACKED EVIDENCE
08

What you receive

Every engagement produces the artefacts your team needs to operate the AIMS — and the evidence pack an accredited certification body expects to review.

01

Gap assessment report

Your current AI governance mapped against ISO/IEC 42001:2023, with a prioritized roadmap to certification-ready.

02

AI risk assessment & treatment plan

A defensible methodology, AI risk register and treatment decisions the auditor can follow.

03

AI system impact assessment

A documented assessment of your AI's consequences for individuals, groups and society, retained per the standard.

04

Statement of Applicability

Each of the 38 Annex A controls, included or excluded, with documented justification.

38 CONTROLS
05

AI policy & procedure suite

The AIMS documentation set, written to the standard and tailored to how you actually develop and use AI.

06

Internal audit report

Independent verification, by CISA-certified IS auditors, that controls are operating, not just documented.

07

Management review pack

The oversight evidence leadership needs for Clause 9.

08

Pre-assessment / mock audit report

A dry run of Stage 1 and Stage 2, with findings to close before the real thing.

09

AI adversarial testing & technical control validation report

Pentest-backed evidence that key controls and the governed AI systems genuinely withstand attack, with remediation guidance and retest.

PENTEST-BACKED
10

Audit-ready evidence pack

Collected, organized artefacts that map cleanly to the controls and clauses.

ISO/IEC 42001:2023 Annex A — 38 controls, 9 objectives CERT-In Empaneled Integrates with ISO 27001 and ISO 27701
09

Why teams choose Intect for ISO 42001

CERT-In Empaneled AI-security depth · offensive heritage

CERT-In Empaneled — An assessor accredited under India's national cybersecurity authority, with CISA-certified IS auditors in-house — credibility the audit room recognizes.

We govern and we break — Most consultancies only write AI policy. We implement the AIMS and then attack the AI it governs, so your evidence is proven, not asserted.

Genuine AI-security depth — We run dedicated AI/ML and autonomous-agent penetration testing. The threats ISO 42001 asks you to manage are threats we test for a living.

One partner, the full cycle — From first gap assessment through Stage 2 support, surveillance and recertification — and integration with your existing ISMS — continuity instead of handoffs.

10

Frequently asked questions

Does Intect issue the ISO 42001 certificate?

No — and no consultancy can. The certificate is issued by an independent, accredited certification body after it conducts the Stage 1 and Stage 2 audits. Intect makes you certification-ready: we build and validate the AIMS, run the AI risk and impact assessments, conduct the internal audit and mock audit, and support you through the certification body's audits. Keeping readiness and certification separate is exactly what preserves the audit's independence — and its value.

How is ISO 42001 different from the EU AI Act?

The EU AI Act is law; ISO 42001 is a voluntary, certifiable management standard. They are complementary — a certified AIMS is a recognized, auditable way to demonstrate the responsible-governance practices regulations increasingly expect. ISO 42001 won't make you automatically compliant with any specific regulation, but it gives you the management system most of them assume you already have. We scope your AIMS with your regulatory obligations in view.

We already have ISO 27001 — do we start from scratch?

No. ISO 42001 uses the same harmonized high-level structure as ISO 27001, so your context, leadership, planning, support and improvement machinery largely carries over. We extend your existing ISMS to AI governance — reusing what works and adding only what is genuinely AI-specific, like the AI impact assessment and the Annex A controls.

What is an AI system impact assessment, and how is it different from a risk assessment?

ISO 42001 requires both. The AI risk assessment looks at risk to the organization and its objectives. The AI system impact assessment looks outward — at the consequences of your AI for the individuals, groups and societies it affects. Both are mandatory clause requirements, and both are documents the auditor will examine. We help you produce each properly.

How many controls are in ISO 42001?

Annex A lists 38 controls organized under 9 control objectives (A.2 to A.10), covering AI policy, internal organization, resources, impact assessment, the AI system life cycle, data, information for interested parties, use of AI, and third-party relationships. Implementing every control is not mandatory — but your Statement of Applicability must justify each inclusion and exclusion. Annex B gives implementation guidance for all of them.

How long does certification take?

It depends on the size and complexity of your AI estate and how mature your current governance is — which is exactly what the gap assessment establishes first. We give you a realistic, evidence-based timeline up front rather than a generic promise, and we sequence the work so nothing blocks the certification audit.

What happens after we're certified?

Certification runs on a three-year cycle. The accredited body conducts annual surveillance audits to confirm the AIMS is still operating, and a full recertification audit at the end of the cycle. This matters especially for AI, because the systems keep learning and changing — we support you across the whole cycle so the certificate stays live and the governance keeps pace.

11

Govern it once, prove it everywhere

A well-built AIMS is stronger when the AI it governs has actually been tested — and when it sits on a mature security and privacy foundation.

Scope an engagement

Build an AI management system that's ready to be certified.

Tell us where you are — first AIMS, an extension of an existing ISO 27001 ISMS, or maintaining an existing certificate — and we'll scope a readiness assessment that fits, or connect you with an assessor.