Key facts
- Published: December 2023 by ISO and IEC, developed by the joint AI committee ISO/IEC JTC 1/SC 42. It is the first AI standard against which an organization can certify its management system.
- Who it is for: Any organization that develops, provides, or uses AI-based products or services, in any sector. It is technology-neutral and applies to any AI system.
- What it addresses: Ungoverned AI risk: unclear accountability, unassessed impacts on people and society, weak data and lifecycle controls, and unmanaged third-party AI.
- How certification works: Accredited certification bodies audit against it. Since July 2025, ISO/IEC 42006:2025 sets the requirements those bodies must meet.
- EU AI Act status: ISO 42001 is not a harmonized standard under the EU AI Act and gives no presumption of conformity. The first AI Act harmonized standard, EN 18286 on quality management, was cited in the Official Journal on October 14, 2026.
How ISO 42001 Works
ISO 42001 asks an organization to build and run a documented AI management system, then prove through an audit that it works. It uses the same Harmonized Structure (formerly Annex SL) as ISO/IEC 27001 and ISO 9001, and follows a Plan-Do-Check-Act cycle. Most organizations follow this path:
- Define scope and context. Decide which AI systems, business units, and interested parties the AIMS covers, and whether the organization acts as an AI developer, provider, user, or several of these.
- Secure leadership commitment. Set an AI policy, assign accountability, and allocate resources.
- Assess AI risks and impacts. Identify risks, assess the impact of each AI system on individuals and society, select the controls that apply, and record the decision in a Statement of Applicability.
- Implement controls. Implement operational controls, competence, and documented information across the AI lifecycle.
- Monitor, audit, and review. Run internal audits and management reviews to check that the system performs as intended.
- Certify and improve. Undergo an external audit by an accredited certification body, then keep improving the system.
Certification follows the usual pattern for ISO management systems: a Stage 1 audit reviews documentation and readiness, and a Stage 2 audit tests whether the system works in practice. Certificates typically run for three years, with surveillance audits in between.
Structure of ISO 42001
The standard has a main body of requirements and four annexes.
Management System Clauses (4–10)
The mandatory requirements an auditor certifies against: context of the organization (4), leadership (5), planning (6), support (7), operation (8), performance evaluation (9), and improvement (10).
Annex A: Reference Controls
Annex A sets out 38 controls grouped under nine control objectives. Organizations select the ones that apply to their AI risks and justify any exclusions.
| Control objective | What it covers | |
| A.2 | Policies related to AI | An AI policy aligned with other organizational policies |
| A.3 | Internal organization | Roles, responsibilities, and reporting of concerns |
| A.4 | Resources for AI systems | Data, tooling, computing, and human resources |
| A.5 | Assessing impacts of AI systems | Impact assessment for individuals, groups, and society |
| A.6 | AI system life cycle | Design, development, verification, deployment, and operation |
| A.7 | Data for AI systems | Data quality, provenance, and preparation |
| A.8 | Information for interested parties | Documentation and information for users and others |
| A.9 | Use of AI systems | Responsible, intended use of AI systems |
| A.10 | Third-party and customer relationships | Suppliers, partners, and customers across the AI supply chain |
Annex B: Implementation Guidance
Guidance on applying each Annex A control, to help organizations decide how to implement it.
Annexes C and D: Objectives, Risk Sources, and Sector Use
Reference material on AI-related organizational objectives and risk sources (Annex C), and on applying the AIMS across domains and sectors (Annex D).
Companion Standards
ISO 42001 sits within a wider family of AI standards. ISO/IEC 23894:2023 gives guidance on AI risk management, ISO/IEC 42005:2025 covers AI system impact assessment, and ISO/IEC 42006:2025 sets requirements for the bodies that audit and certify an AIMS.
ISO 42001 vs Related Standards and Regulations
People often confuse ISO 42001 with other AI and security frameworks. The main difference is what each one governs and whether it is binding.
| ISO 42001 | ISO/IEC 27001 | NIST AI RMF | EU AI Act | |
| Governs | An organization’s AI management system | An organization’s information security management system | How organizations identify and manage AI risk | AI systems and GPAI models placed on the EU market |
| Nature | Voluntary international standard | Voluntary international standard | Voluntary U.S. framework | Binding EU regulation |
| Certifiable | Yes, by accredited bodies | Yes, by accredited bodies | No | No organizational certificate; high-risk systems undergo conformity assessment |
| Core question | “Do we govern AI responsibly, and can we prove it?” | “Do we manage information security systematically?” | “How do we identify and manage AI risk?” | “Does this AI system comply with the law?” |
ISO 42001 certification does not equal EU AI Act compliance. The standard audits an organizational management system, while the Act sets legal duties for specific AI systems. ISO 42001 can support readiness for obligations such as risk management, documentation, and human oversight, but it does not replace conformity with the Act itself.
ISO 42001 and EN 18286
EN 18286:2026 is the quality management system standard written for the EU AI Act. Since its citation in the Official Journal on October 14, 2026, providers that apply it gain a presumption of conformity with the Article 17 quality management requirements it covers. It is designed to sit on top of existing management systems such as ISO 42001, not to replace them. A mature ISO 42001 program is a strong head start, but organizations still need a gap analysis against EN 18286.
Why ISO 42001 Matters
- Trust and procurement. Enterprise buyers increasingly ask AI vendors for independent evidence of responsible AI governance, and an accredited certificate is third-party proof.
- Regulatory readiness. A structured AIMS gives obligations such as risk management, documentation, human oversight, and transparency a management system to live in, even though it does not satisfy the law on its own.
- Reuse of existing systems. Because it shares a structure with ISO/IEC 27001, organizations with an information security management system can reuse parts of their scope, risk process, and audit routines.
- Risk of overclaiming. “ISO 42001 certified” and “EU AI Act compliant” are different statements. Treating one as the other creates legal and reputational exposure.
Where ISO 42001 Meets the Browser
An AIMS covers the AI lifecycle, but many AI systems reach users through web applications. Chatbots, personalization engines, and embedded AI SDKs read form fields and page content in the browser, then send that context to external models. ISO 42001 is technology-neutral and doesn’t prescribe browser controls, so organizations have to identify and treat this layer in their own risk assessment. Several Annex A areas reach it directly:
- Data for AI systems (A.7). AI inputs are often assembled in the browser. Without controls, AI-powered scripts can read sensitive fields that were never meant to enter a model.
- Third-party relationships (A.10). Third-party AI widgets and SDKs run in your pages, and vendors can add fourth-party AI services without notice.
- AI system life cycle (A.6). Scoring logic, prompts, model configuration, or API keys placed in client-side code can be inspected, extracted, or tampered with after deployment.
- Evidence for auditors. Without a runtime record, it’s hard to show which AI scripts ran and what data they accessed during live sessions.
Best Practices for ISO 42001 Implementation
- Scope deliberately. Define which AI systems and business units the AIMS covers, and revisit the scope as your AI use grows.
- Build on ISO/IEC 27001. Reuse your existing management system structure, risk process, and internal audit routines wherever possible.
- Inventory your AI systems. Record every model, use case, and third-party AI service, including AI-powered scripts embedded in web applications.
- Run AI impact assessments. Assess how each system could affect individuals and society, and feed the results into risk treatment. ISO/IEC 42005 gives detailed guidance.
- Treat third-party AI as in scope. Apply the third-party controls to vendors, SDKs, and AI widgets that run in your pages, not only to contracted suppliers.
- Include the browser in your risk assessment. Account for exposed application logic, secrets in front-end code, and what data AI-powered scripts can read and send.
- Map to the EU AI Act. Link your controls to the Act’s obligations, run a gap analysis against EN 18286 if you provide high-risk AI systems, and keep legal analysis separate from certification evidence.
- Audit and improve continuously. Run internal audits and management reviews on a schedule, not only before the certification audit.
- Choose an accredited certification body. Check that the body is accredited for ISO 42001 and works to ISO/IEC 42006.
Solutions by Jscrambler
ISO 42001 governs AI at the organizational level. Much of the AI risk it asks you to treat takes shape in the browser, beyond the traditional security edge, where AI inputs are assembled, and third-party scripts execute. The Jscrambler Client-Side Security Platform enforces policy at that point of creation.
- AI Data Governance & Security: Discovers AI-powered scripts and agents running in the browser and maps the data they access. Least-privilege runtime controls stop them from reading sensitive fields and block unauthorized transmission of prompts and context, with audit-ready activity logs that can support AIMS evidence.
- LLM-Resilient Code Protection: Jscrambler Code Integrity protects first-party application logic, including AI-related scoring and decisioning logic, with polymorphic obfuscation and anti-tampering, reducing the risk of inspection, extraction, and unauthorized modification.
These solutions support risk treatment and evidence within an AI management system. They do not make an organization ISO 42001 certified or compliant with any regulation on their own.
FAQs
What is ISO 42001?
ISO 42001, formally ISO/IEC 42001:2023, is the first international standard for an AI management system. It sets requirements for governing AI responsibly through policies, risk and impact assessment, controls, monitoring, and continual improvement.
Is ISO 42001 mandatory?
No. It is a voluntary standard. Organizations can choose to implement it and seek certification from an accredited body to show their AI governance has been independently audited.
Does ISO 42001 certification make me compliant with the EU AI Act?
No. The EU AI Act is binding law that imposes obligations on specific AI systems. ISO 42001 is not a harmonized standard under the Act, so it does not create a presumption of conformity. The first harmonized standard, EN 18286 (cited October 14, 2026), covers quality management for providers and builds on systems such as ISO 42001.
How is ISO 42001 different from ISO/IEC 27001?
ISO/IEC 27001 governs information security management, while ISO 42001 governs how an organization develops, provides, and uses AI. They share the same management-system structure, so organizations with ISO/IEC 27001 can often reuse parts of it.
How many controls are in ISO 42001?
Annex A lists 38 reference controls under nine control objectives, from AI policies (A.2) to third-party and customer relationships (A.10).
How long is ISO 42001 certification valid?
Like other ISO management system certificates, it typically runs for three years, with surveillance audits in between and a recertification audit at the end of the cycle.
Who should pursue ISO 42001 certification?
Any organization that develops, provides, or uses AI-based products or services, particularly AI vendors that need to show enterprise customers independent evidence of responsible governance.
Does ISO 42001 cover third-party AI and client-side code?
It covers third-party relationships through Annex A and is technology-neutral. It does not prescribe browser-specific controls, so organizations should include browser-delivered AI, third-party AI scripts, and client-side application logic in their own risk assessment.
Related Resources