Cybernion
← Insights

ISO 27090

ISO 27090 vs ISO 42001

26 Oct 2026 · Updated 26 Oct 2026

ISO 42001 and ISO 27090 both address AI, but they do different jobs. ISO/IEC 42001:2023 is a certifiable AI management system standard, the governance wrapper for how an organisation builds and uses AI. ISO/IEC 27090:2026 is guidance on the security threats specific to AI and how to mitigate them. You use 27090 to inform the risk work that 42001 requires.

What is the core difference?

One is a management system you certify, the other is threat guidance you apply. ISO 42001 is a requirements standard: it sets out how to run an AI management system, with leadership, policies, an AI risk and impact assessment, human oversight and continual improvement, on the same clause 4 to 10 backbone as ISO 27001. An accredited body can audit and certify it. ISO 27090 is a guidance standard: it describes the security threats specific to AI systems, such as data poisoning, model theft and prompt injection, and how to detect and treat them across the life cycle. It has no “shall” clauses and no certificate. In short, 42001 governs the AI, and 27090 explains the attacks that AI invites.

Is either one certifiable?

Only ISO 42001. It can be certified by an accredited certification body on a three year cycle of a Stage 1 and Stage 2 audit followed by surveillance audits, the same pattern as ISO 27001. ISO 27090 cannot be certified, because guidance standards describe good practice rather than auditable requirements. If a tender or customer asks for proof of responsible AI, the credential they can actually verify is ISO 42001 certification. ISO 27090 strengthens the work behind that certificate, but it is not itself a claim you can be audited against.

How do they fit together?

ISO 27090 feeds the risk engine inside ISO 42001. ISO 42001 requires an AI risk assessment and an AI system impact assessment, but it does not list the technical attacks an AI system faces. That is exactly what ISO 27090 supplies. A team running an AI management system can use 27090 as the threat library for the security side of its risk assessment, mapping each relevant compromise type to a control and an owner. The two standards even share lineage: 27090 points to ISO 42001 for governance and to ISO 23894 for AI risk management, while 42001’s Annex A controls on the AI life cycle, data and third parties are where many 27090 mitigations land. See the ISO 42001 Annex A controls for that control set.

ISO 42001 ISO 27090
Type Management system (requirements) Guidance
Certifiable Yes No
Focus Governance of AI end to end Security threats specific to AI
Published December 2023 October 2026
Use it to Build and certify AI governance Assess and treat AI specific threats

Which do I need?

Most organisations serious about AI will use both, in that order. If you need a position you can show a customer or regulator, pursue ISO 42001; it is the certifiable governance standard, and demand for it is commercial and procurement driven. Use ISO 27090 as the reference that makes the security part of that work concrete, especially if you develop models or handle training data. If you are not ready for a full management system, 27090 still stands alone as a practical checklist of AI threats to assess your systems and vendors against. Cybernion helps organisations build and certify AI governance through ISO 42001 readiness.

Frequently asked questions

Is ISO 27090 or ISO 42001 certifiable?

Only ISO 42001. It is a management system standard an accredited body can certify. ISO 27090 is guidance with no certification.

Do I need both?

If you govern AI seriously, yes in effect. ISO 42001 is the system you certify; ISO 27090 is the AI threat reference you assess against inside it.

Which should I start with?

Start with ISO 42001 if you need a certifiable governance position. Use ISO 27090 to inform the AI risk and security work that 42001 requires.


Written by Gaurav Vikash, an ASD endorsed IRAP assessor and senior cyber security leader with 18 years of experience across Australia, the UK and Asia, including CISO and senior security leadership roles. He holds CISSP, CISA, CISM and CRISC and is an ISO 27001 Lead Auditor and ISO 42001 Lead Implementer, and speaks regularly at industry conferences.

If you are weighing up AI governance and AI security, talk to us.

Sources:

  1. ISO/IEC 27090:2026, Cybersecurity, Artificial intelligence, Addressing security threats and compromises to AI systems, October 2026
  2. ISO/IEC 42001:2023, Information technology, Artificial intelligence, Management system, December 2023
  3. ISO/IEC 23894:2023, Artificial intelligence, Guidance on risk management, 2023

Last updated: 26 October, 2026

Talk to us about your engagement

A scoped proposal within one business day.