What ISO/IEC 42001 covers
ISO/IEC 42001:2023, Information technology - Artificial intelligence - Management system, was developed by the joint ISO and IEC committee on artificial intelligence, ISO/IEC JTC 1/SC 42, and published in December 2023. ISO describes it as specifying requirements for establishing, implementing, maintaining and continually improving an artificial intelligence management system (AIMS) within an organisation, and says it is meant for organisations of any size, in any industry, that develop, provide or use AI-based products or services.
It is a management system standard, not a technical specification for a particular model. ISO explains that implementing it means putting policies and procedures in place for governing AI across the organisation using the Plan-Do-Check-Act cycle, rather than prescribing how a specific AI application should work. ISO says many of its management system standards, including ISO/IEC 27001, follow a common Harmonized Structure so that users familiar with one find the next familiar. The ISO/IEC 42001 clause headings listed in the crosswalk below (context, leadership, planning, support, operation, performance evaluation and improvement) follow the same pattern.
The crosswalk listed on NIST's AI Resource Center, mapped against the final draft of the standard, lists clauses of the standard including the organisation's context and the scope of the AI management system, leadership commitment and an AI policy, AI risk assessment, AI risk treatment and AI system impact assessment, competence and awareness, internal audit and management review, and corrective action and continual improvement. It also maps to the standard's AI-specific controls and guidance, including AI roles and responsibilities, documentation of data, tooling and computing resources, data quality and data provenance, verification and validation, operation and monitoring, event logs, information for interested parties, and allocating responsibilities to suppliers and customers.
ISO publishes related guidance standards. ISO/IEC 23894:2023 gives guidance on AI risk management, and ISO/IEC 42005:2025, which ISO describes as complementing ISO/IEC 42001, gives guidance on conducting AI system impact assessments.
Certification versus alignment
ISO states that certification to ISO/IEC 42001 is voluntary, that ISO itself does not certify organisations, and that certification is carried out by independent certification bodies, which may be accredited by national accreditation bodies. ISO/IEC 42006:2025, published in July 2025, sets additional requirements for bodies that audit and certify AI management systems against ISO/IEC 42001, building on ISO/IEC 17021-1, which it supplements with AI-specific requirements.
ISO's guidance on certification also says accreditation is not compulsory, and that a certification issued by an accredited body can be checked through the IAF CertSearch database or by contacting the certification body. If a partner tells you it is certified, ask for the certificate, the certification body and the scope, because a certificate applies to the scope of the AI management system that was audited, which the organisation determines under the standard's scope clause.
Alignment means adopting the standard's requirements, or some of them, without a third-party audit. ISO notes that adopting the standard delivers value whether or not an organisation seeks certification. A statement that a company is 'aligned with ISO/IEC 42001' is a self-description, not an independent finding, and a partner reviewing it may look for the evidence behind it: the AI policy, the inventory of AI systems, the risk and impact assessments and the records of review.
| Certified to ISO/IEC 42001 | Aligned with ISO/IEC 42001 | |
|---|---|---|
| Who decides | An independent certification body, after an audit | The organisation itself |
| What a partner can check | The certificate, the certification body, its accreditation, if any, and the audited scope | Whatever documents the organisation chooses to share |
| Standard for the auditor | ISO/IEC 17021-1, supplemented for AI management systems by ISO/IEC 42006 (ISO notes accreditation is not compulsory) | None |
| Required by ISO | No; ISO says certification is voluntary | No |
How it relates to the NIST AI RMF and ISO/IEC 27001
The NIST AI Risk Management Framework (AI RMF 1.0, NIST AI 100-1) was released on 26 January 2023. NIST describes it as intended for voluntary use and as rights-preserving, non-sector-specific and use-case agnostic. Its core has four functions, Govern, Map, Measure and Manage, broken into categories and subcategories of outcomes. NIST added a Generative AI Profile, NIST AI 600-1, in July 2024, and its AI RMF page says AI RMF 1.0 is being revised as part of the White House AI Action Plan. Where ISO/IEC 42001 sets requirements that a certification body audits against, the AI RMF describes outcomes an organisation can choose to pursue.
The two fit together. The crosswalk from the AI RMF to ISO/IEC 42001 listed on NIST's AI Resource Center was provided by Microsoft, and maps, for example, the AI RMF outcome that executive leadership takes responsibility for AI risk decisions to the standard's leadership and management review clauses, and production monitoring of AI systems to its monitoring, operation and event-log provisions. NIST notes that listing a crosswalk does not imply NIST endorsement of the resource the AI RMF is mapped to, nor that either document fully covers the other.
ISO/IEC 27001:2022 sets requirements for an information security management system. It deals with protecting information. ISO/IEC 42001 applies the same management system approach to the development, provision and use of AI, adding elements such as AI system impact assessment and controls over the data and lifecycle of AI systems. An organisation that already holds ISO/IEC 27001 can reuse its management system machinery, such as document control, internal audit and management review, for an AI management system.
| ISO/IEC 42001:2023 | NIST AI RMF 1.0 | ISO/IEC 27001:2022 | |
|---|---|---|---|
| Publisher | ISO and IEC (JTC 1/SC 42) | NIST, US Department of Commerce | ISO and IEC (JTC 1/SC 27) |
| What it is | Requirements for an AI management system | Voluntary framework of AI risk outcomes | Requirements for an information security management system |
| Structure | Management system clauses plus AI-specific controls and guidance | Govern, Map, Measure and Manage functions | Management system clauses plus information security controls |
| Third-party certification | Available from independent certification bodies; voluntary | None described by NIST; NIST says it is intended for voluntary use | Available from independent certification bodies |
How it relates to GxP quality systems
GxP obligations come from regulation, not from voluntary standards. ISO says ISO/IEC 42001 does not replace laws or regulations and is instead a management framework that helps organisations meet their compliance obligations. A biotech that uses an AI system in a GMP, GLP or GCP process still has to meet the applicable rules, for example 21 CFR Part 11 for electronic records and electronic signatures in the United States where the system creates or maintains records required by FDA regulations, and to validate the system for its intended use. A 42001 certificate is not a substitute for that validation, although a supplier's certificate may inform a risk-based supplier assessment.
The pharmaceutical quality system guideline ICH Q10, published by FDA as guidance, describes a model for a pharmaceutical quality system that is based on ISO quality concepts and includes applicable GMP regulations. ICH Q10 states that it is not intended to create new expectations beyond current regulatory requirements. Neither ICH Q10 nor Part 11 mentions ISO/IEC 42001; both predate it. For medical devices, including software that is itself a medical device, ISO's quality management system standard is ISO 13485:2016, which sets requirements for quality management systems in the design and manufacture of medical devices. In the United States, FDA's Quality Management System Regulation, which amended 21 CFR Part 820 and became effective on 2 February 2026, incorporates ISO 13485:2016 by reference, so for finished device manufacturers subject to the QMSR it is part of a regulatory requirement rather than a voluntary choice.
In practice the overlap is in the management machinery. A quality system already has document control, change control, training records, deviations and corrective action, internal audit and management review. An AI management system needs most of the same things for AI. Extending existing quality and security processes to cover AI can be simpler than building a separate system beside them, and keeps one owner for each record.
When a sponsor or partner may ask for it
ISO says certification is voluntary, and none of the GxP texts cited on this page require it. Where the question comes up for a biotech, it is more likely to come through a contract or due-diligence review than from a regulator. Situations where it may come up include:
A biotech or platform company that sells AI-based discovery, analysis or software services to pharma companies, and whose customer's vendor assessment asks how AI risk is governed. A CRO or service provider that uses AI agents or models on a sponsor's data, where the sponsor wants evidence of AI policy, data handling and human oversight. A partnership, licensing or financing diligence process in which the counterparty asks for an AI governance description alongside security certifications such as ISO/IEC 27001. A company that develops AI systems it intends to place on regulated markets and wants an external audit of its AI management system.
If you mainly use AI tools inside your own research and are not selling AI-based services, a certificate may not be the first need. An inventory of AI systems, a written AI policy, risk and impact assessments for the systems that matter, and clear ownership can answer many partner questions, and are also the first steps toward the standard if you later decide to certify.
Deciding whether to certify
Start by writing down who is asking and what they will accept. If a customer's contract or vendor assessment names ISO/IEC 42001 certification, certification may be the shortest path. If they ask how AI risk is managed, a documented management system aligned with the standard, or with the NIST AI RMF, may be enough. Decide the scope before talking to certification bodies: which AI systems, which sites and which activities. Ask any certification body you consider whether it is accredited for ISO/IEC 42001 and whether it works to ISO/IEC 42006. Keep GxP validation, data protection and security obligations on their own tracks; certification does not discharge them. This page summarises public ISO and NIST material; it is not legal or regulatory advice.
Whether or not you pursue a certificate, the starting point is knowing which AI systems you use and what data they touch. Atlas Bio's Secure AI for Life Sciences begins with an AI risk assessment of current AI use and data exposure, a starting point for the inventory and risk assessments described above; the assessment tooling is built and tested on synthetic data only, not yet used in a client assessment. Status, October 2026. In use in Atlas Bio's own agent platform today: a separate verification step that checks agent work, and a hold that stops a task until a person decides. Built and tested on synthetic data, not yet used with a client: the AI risk assessment tooling (AI discovery, data-flow and vendor-terms mapping, and a NIST AI RMF gap report) and a record of what data an AI used, what it produced and who reviewed it. In development: a customer-controlled AI environment in your own Azure tenant, an AI gateway, agent permissions, evidence linkage and notifications. Each engagement states which controls are built. Atlas Bio is not a certification body and does not issue ISO certificates. Decisions on certification scope, GxP scope and validation stay with you, under your quality system.
Is ISO/IEC 42001 certification mandatory?
No. ISO states that certification to ISO/IEC 42001 is voluntary. An organisation might seek it to give customers or partners an independent audit of its AI management system against the standard.
Who issues ISO/IEC 42001 certificates?
Independent certification bodies, not ISO. ISO says it does not certify organisations. ISO/IEC 42006:2025 sets additional requirements for bodies that audit and certify AI management systems against ISO/IEC 42001.
Is ISO/IEC 42001 the same as the NIST AI RMF?
No. ISO/IEC 42001 sets requirements for an AI management system that a certification body can audit. The NIST AI RMF is a voluntary US framework organised into Govern, Map, Measure and Manage functions. NIST's AI Resource Center lists a crosswalk between the two, and NIST notes that listing a crosswalk does not imply that either document comprehensively covers the other.
Does ISO/IEC 42001 replace GxP computer system validation?
No. ISO says the standard does not replace laws or regulations. AI systems used in GxP processes still need to meet the applicable rules, such as 21 CFR Part 11 in the United States where they create or maintain records required by FDA regulations, and be validated for their intended use.
Do we need ISO/IEC 27001 before ISO/IEC 42001?
ISO publishes them as separate standards, and ISO's description of ISO/IEC 42001 does not list ISO/IEC 27001 as a prerequisite. They cover different subjects, information security and AI management, but ISO says many of its management system standards follow a common Harmonized Structure, and ISO/IEC 42001's clause headings follow the same pattern, so an existing ISO/IEC 27001 programme can give you document control, internal audit and management review processes to reuse.
- ISO: ISO/IEC 42001:2023 - Information technology - Artificial intelligence - Management system
- ISO: ISO/IEC 42001 explained
- ISO: Certification (conformity assessment)
- ISO: ISO/IEC 42006:2025 - Requirements for bodies providing audit and certification of AI management systems
- ISO: Management system standards and the Harmonized Structure
- ISO: ISO/IEC 27001:2022 - Information security management systems - Requirements
- ISO: ISO/IEC 23894:2023 - Artificial intelligence - Guidance on risk management
- ISO: ISO/IEC 42005:2025 - Artificial intelligence - AI system impact assessment
- ISO: ISO 13485:2016 - Medical devices - Quality management systems
- NIST: AI Risk Management Framework
- NIST AI 100-1: Artificial Intelligence Risk Management Framework (AI RMF 1.0), January 2023
- NIST AI 600-1: AI RMF Generative Artificial Intelligence Profile, July 2024
- NIST AI Resource Center: Crosswalk documents
- NIST AI Resource Center: NIST AI RMF to ISO/IEC FDIS 42001 crosswalk (provided by Microsoft)
- FDA: Quality Management System Regulation (QMSR), 21 CFR Part 820
- FDA: Q10 Pharmaceutical Quality System (ICH guidance)
- eCFR: 21 CFR Part 11 - Electronic Records; Electronic Signatures
Talk to Atlas Bio.
Sponsors, researchers and clinicians: tell us the decision you are trying to make and we will show what the platform can and cannot answer.
Contact Us →