ISO/IEC 42001 matters to UK businesses that build, sell or embed AI in their products, especially if enterprise customers or EU buyers ask how they govern AI. It is voluntary, and no UK law requires it. For organisations that only use off-the-shelf AI tools, a proportionate AI policy and risk assessment usually comes first, with certification later if customers ask.
ISO/IEC 42001 is the first international management system standard for artificial intelligence. It has moved quickly from a niche topic to a line in procurement questionnaires, helped by the EU AI Act and growing customer scrutiny of AI products. This article explains what the standard is, who it genuinely matters for, and what it does and does not do.
Executive Summary & Key Takeaways
- ISO/IEC 42001:2023 sets requirements for an AI management system (AIMS). It is voluntary and applies to organisations of any size that develop, provide or use AI.
- Accredited certification is now available in the UK. UKAS accredits certification bodies for ISO/IEC 42001, supported by ISO/IEC 42006:2025, which sets requirements for the auditors themselves.
- ISO/IEC 42001 is not a harmonised standard under the EU AI Act, so certification does not by itself give a presumption of conformity with the Act.
- The standard governs how you manage AI. It does not test whether your models resist prompt injection or data leakage; that needs security testing.
- If you already run ISO 27001, much of the management system structure carries across.
What ISO/IEC 42001 is
ISO/IEC 42001:2023 was published in December 2023. It specifies requirements for establishing, implementing, maintaining and continually improving an AI management system within an organisation, and is designed for organisations providing or using AI-based products or services (ISO, retrieved 10 November 2026).
Like ISO 27001, it is a management system standard. Rather than prescribing technical details for individual AI applications, it asks you to set AI policy and objectives, put processes in place to meet them, assess and treat AI-related risks, and review and improve the system using the Plan-Do-Check-Act cycle (ISO, retrieved 10 November 2026). It sits alongside other ISO AI standards, such as ISO/IEC 23894 on AI risk management guidance, but 42001 is the one you can certify against.
Certification in the UK
Certification is carried out by an independent certification body. In July 2025 ISO and IEC published ISO/IEC 42006, which sets additional requirements for bodies that audit and certify AI management systems, building on ISO/IEC 17021-1 (ISO, retrieved 10 November 2026). UKAS has completed its pilot accreditation programme for ISO/IEC 42001 and now accredits certification bodies to certify against it, with BSI the first to be accredited (UKAS, retrieved 10 November 2026).
Falx is not a certification body and cannot issue ISO certificates. We help organisations prepare: AI system inventories, risk assessments, policies and the security testing that supports them.
Why it is getting attention in the UK
Three pressures are pushing ISO/IEC 42001 up the agenda.
Customer due diligence. Enterprise buyers increasingly ask suppliers how they govern AI. The Cyber Security Breaches Survey 2025/2026 found that 31% of UK businesses were using, adopting or considering AI, but only 24% of that group reported having cyber security practices in place to manage AI risks (DSIT and Home Office, retrieved 10 November 2026). Buyers know that gap exists, and a recognised standard is an easy way to ask about it.
The EU AI Act. The AI Act applies to UK organisations that place AI systems on the EU market or whose AI output is used in the EU (Regulation (EU) 2024/1689, Article 2, AI Act Service Desk, retrieved 10 November 2026). Its timetable has moved. The Digital Omnibus on AI, Regulation (EU) 2026/1744, entered into force on 27 July 2026 and delayed the high-risk obligations: they now apply from 2 December 2027 for stand-alone high-risk systems listed in Annex III, and from 2 August 2028 for high-risk AI embedded in regulated products (European Commission, retrieved 10 November 2026).
On 14 October 2026 the Commission published the reference to the first harmonised standard under the AI Act, EN 18286:2026, covering the quality management system that high-risk AI providers need under Article 17 (EUR-Lex, Implementing Decision (EU) 2026/2274, retrieved 10 November 2026). ISO/IEC 42001 is not on that list. An existing AI management system is a useful foundation for the quality management work the Act requires, but holding a 42001 certificate does not on its own demonstrate AI Act conformity.
UK guidance. The UK has not legislated an AI-specific regime equivalent to the EU AI Act. On 12 October 2026 the government said it is considering how the Cyber Security and Resilience Bill could better address AI-related risks (Hansard, retrieved 10 November 2026). In the meantime, the voluntary Code of Practice for the Cyber Security of AI, published by DSIT in January 2025, sets 13 principles across the AI lifecycle and became the basis for ETSI EN 304 223 (GOV.UK and ETSI, retrieved 10 November 2026).
Who it matters for
| model | Does ISO/IEC 42001 matter? |
|---|---|
| You build AI products or features and sell them to enterprises | Yes, increasingly. Expect AI governance questions in due diligence, and consider certification if buyers ask for it. |
| You provide high-risk AI systems into the EU | Yes, as a governance foundation, but plan for EN 18286 and the AI Act’s own requirements alongside it. |
| You fine-tune or integrate third-party models into a customer-facing service | Often. The standard helps you show control over models, data and suppliers. |
| You are regulated and use AI in decisions about customers, such as credit or claims | Worth considering as a framework, alongside your regulator’s expectations and UK GDPR rules on automated decisions. |
| You only use off-the-shelf tools such as AI assistants | Usually not yet. An AI acceptable use policy, a register of tools and a risk assessment are more proportionate. |
What ISO/IEC 42001 does not do
The standard is valuable, but it is easy to expect too much of a certificate.
- It does not test your AI’s security. A management system shows you have processes for identifying and treating risk. It does not show that a chatbot resists prompt injection or that a retrieval pipeline cannot leak another customer’s data. Our guide to AI security explains the threats that need hands-on testing.
- It does not replace data protection duties. AI that processes personal data still needs a lawful basis, transparency and, where processing is high risk, a Data Protection Impact Assessment under UK GDPR.
- It does not equal AI Act compliance. As above, it is not a harmonised standard.
- It is not a substitute for ISO 27001. It governs AI. Your wider information security still needs its own management.
How it fits with ISO 27001 and other frameworks
If you already hold ISO 27001, you have a head start. Both are management system standards, so scope, risk assessment, internal audit, management review and continual improvement work in a familiar way, and the two can be run as one integrated system. The AI-specific work covers questions an ISMS does not, such as the ethical impact of AI systems, transparency and traceability, and how models that keep learning are managed over time.
Other frameworks can sit alongside it. The NIST AI Risk Management Framework is a common reference for risk practice, and the UK Code of Practice gives a security baseline for AI systems. Our AI and ML security service maps governance work to the frameworks your customers recognise, so one set of evidence serves several of them.
What drives effort
Effort depends on how many AI systems are in scope, whether you develop models or only integrate them, how mature your existing management systems are, and how much AI-related documentation already exists. An organisation with ISO 27001 and a clear AI system register will need far less work than one starting from scratch. We do not publish prices; certification body fees are quoted by the body you choose.
Common mistakes
- Pursuing certification before knowing what AI you run. Start with an inventory of AI systems and tools, who owns them and what data they touch.
- Assuming the certificate proves your AI is secure. Pair governance with testing against threats such as prompt injection and data leakage.
- Treating 42001 as EU AI Act compliance. Map your obligations under the Act separately.
- Running AI governance in isolation. Build it into your existing risk register and ISO 27001 processes rather than creating a parallel system.
- Scoping every AI tool equally. Focus effort on the systems that affect customers, decisions or sensitive data.
Frequently asked questions
Do we need ISO 27001 before ISO 42001?
No, the standards are independent and you can certify to ISO 42001 alone. In practice, an existing ISMS makes the AI management system faster to build, because the management system structure is the same. You can also integrate the two under a single scope and audit programme.
We only use tools like AI assistants. Is ISO 42001 relevant?
Rarely as a certification target. What matters more is knowing which tools are in use, what data reaches them and what staff are allowed to do with them. A short policy, a tool register and a risk assessment cover most of the governance value at a fraction of the effort.
How do we check that a supplier’s ISO 42001 certificate is credible?
Check that the certification body is accredited for ISO/IEC 42001 by UKAS or another accreditation body that is a signatory to the International Accreditation Forum’s recognition arrangement. Then check the certificate’s scope covers the AI product or service you are buying, and that it is current.
Will ISO 42001 certification help with UK GDPR?
It can support it. An AI management system produces records such as impact assessments, data documentation and oversight procedures that also help demonstrate accountability under UK GDPR. It does not replace a Data Protection Impact Assessment where one is required.
Getting started
The most useful first step is an inventory: which AI systems you build or use, who owns them, what data they process and who relies on their outputs. That shows whether ISO/IEC 42001 is worth pursuing now and where testing is needed first. Our AI and ML security team can help with both governance and testing. Talk to our team to discuss where to start.

Ashley Argyle
Writing about cyber security and technology at Falx.





