Only EN 18286 gives you a presumption of conformity under the EU AI Act. ISO 42001 does not. On 14 October 2026, EN 18286:2026 became the first harmonised standard cited in the Official Journal under the AI Act, covering the quality management system that providers of high-risk AI must operate under Article 17. ISO 42001 remains valuable as an organisation-wide AI governance system, and EN 18286 is designed to sit on top of it, but an ISO 42001 certificate alone proves nothing in law.

This guide explains what each standard covers, what the presumption of conformity actually gives you (and what it doesn't), how much of your ISO 42001 work carries over, and what providers of high-risk AI should do next.

EN 18286:2026
Artificial intelligence — Quality management system for EU AI Act regulatory purposes (CEN-CENELEC)
Approved by CEN
12 July 2026
Cited in the Official Journal
14 October 2026, by Commission Implementing Decision (EU) 2026/2274
Presumption covers
Article 17(1) quality management system and part of Article 11(1) technical documentation
ISO/IEC 42001:2023
International AI management system standard; certifiable, but no presumption of conformity
Who EN 18286 is for
Providers of high-risk AI systems, including those outside the EU
High-risk deadline
2 December 2027 (Annex III); 2 August 2028 (Annex I)

What "presumption of conformity" actually means

Under Article 40 of the AI Act, a high-risk AI system that conforms to a harmonised standard, or the relevant parts of one, whose reference is published in the Official Journal is presumed to meet the AI Act requirements that standard covers. In practice, the burden shifts. You show you followed the standard, and a regulator must then show why that isn't enough.

Three conditions must all be true:

  • The standard is a European harmonised standard, developed by CEN, CENELEC or ETSI at the Commission's request
  • Its reference is published in the Official Journal of the EU. Publication by CEN-CENELEC alone is not enough.
  • The presumption applies only to the specific requirements the standard covers, within any limits set in the citation

EN 18286 meets all three for quality management. ISO 42001 meets none of them. It is an international ISO/IEC standard, it was not developed under the Commission's standardisation request, and it is not cited in the Official Journal.

An ISO 42001 certificate is not AI Act conformity

Some vendors still market ISO 42001 as "EU AI Act certification". It isn't. The Commission has been clear that adhering to ISO 42001 or ISO 9001 on its own does not confer a presumption of conformity with Article 17.

EN 18286 vs ISO 42001 at a glance

DimensionEN 18286:2026ISO/IEC 42001:2023
TypeEuropean harmonised standardInternational management system standard
Developed byCEN-CENELEC JTC 21, at the Commission's requestISO/IEC JTC 1/SC 42
PurposeMeet the AI Act's quality management obligationGovern AI responsibly against the organisation's own objectives
Who it is forProviders of high-risk AI systemsAny organisation that develops, provides or uses AI
FocusRegulatory compliance and protection of health, safety and fundamental rightsOrganisational AI objectives, risk and responsible use
Presumption of conformityYes, for Art. 17(1) and part of Art. 11(1)No
CertifiableNo certification scheme of its own; used in conformity assessmentYes, by accredited certification bodies
AI system impact assessmentNo dedicated clauseYes (Clauses 6.1.4 and 8.4)
Serious-incident and non-compliance proceduresYes, tied to the AI ActNot tied to AI Act timelines
Technical documentation for Annex IVYesNo
StructureSame high-level management-system structure, layered on ISO 42001 or ISO 9001Clauses 4–10 plus Annex A controls

What EN 18286 covers: the 13 elements of Article 17

Article 17(1) lists 13 things a high-risk provider's quality management system must include, labelled (a) to (m). EN 18286 turns them into auditable requirements.

Art. 17(1)RequirementTypically covered by ISO 42001?
(a)Regulatory compliance strategy, including conformity assessment and change managementNo
(b)Design, design control and design verification techniquesPartly
(c)Development, quality control and quality assurancePartly
(d)Examination, test and validation proceduresPartly
(e)Technical specifications and standards appliedNo
(f)Data management systems and proceduresLargely
(g)The risk management system under Article 9Partly
(h)Post-market monitoring under Article 72Partly
(i)Serious-incident reporting under Article 73No
(j)Communication with authorities, notified bodies and othersNo
(k)Record-keeping of all relevant documentationLargely
(l)Resource management, including security of supplyPartly
(m)An accountability framework for management and staffLargely

"Typically covered" is a practical assessment of a well-run ISO 42001 system, not an official mapping. Your own gap analysis should be done clause by clause.

13quality management elements required by Article 17(1)
1stharmonised standard cited under the AI Act (EN 18286, 14 Oct 2026)
10areas in the Commission's AI Act standardisation request

What EN 18286 does NOT give you

This is the part most early commentary gets wrong. The presumption is real, but narrow.

It covers the quality system, not the AI system

EN 18286 gives a presumption for how you manage quality. It does not presume your system meets the other high-risk requirements: risk management (Art. 9), data governance (Art. 10), record-keeping (Art. 12), transparency (Art. 13), human oversight (Art. 14), and accuracy, robustness and cybersecurity (Art. 15).

The other standards aren't cited yet

The Commission's standardisation request covers 10 areas. As of late October 2026, EN 18286 is the only one cited in the Official Journal. Further references will be added as other standards are assessed and accepted.

Biometric systems still need a notified body for now

For remote biometric systems under Annex III, point 1, self-assessment is only available if harmonised standards or common specifications cover all the essential requirements. With only EN 18286 cited, that condition isn't met. Most other Annex III systems (points 2 to 8) use internal control, whatever the standards situation.

Working out where your quality system stands?

I share practical breakdowns of AI Act standards and ISO 42001 most weeks. Connect with me on LinkedIn and tell me whether you're starting from ISO 42001, ISO 9001 or neither.

How much of your ISO 42001 work carries over

A lot, at the management-system level. EN 18286 uses the same high-level structure as ISO 42001 and ISO 9001, and the Commission describes it as designed to be layered on top of those standards rather than replace them. It also includes an informative annex relating its clauses to ISO 42001.

What ISO 42001 already gives you

  • Leadership commitment, policy and roles
  • A risk-based planning process
  • Document control and record keeping
  • Data management for AI
  • Internal audit and management review
  • Supplier management for AI components

What EN 18286 adds for the AI Act

  • A regulatory compliance strategy for each high-risk system
  • Annex IV technical documentation
  • Change management for systems that keep learning
  • Article 73 serious-incident reporting procedures
  • Communication with authorities and notified bodies
  • Post-market monitoring tied to Article 72

There is also a difference in mindset. ISO 42001 asks you to manage risks against your organisation's objectives. EN 18286 ties quality to regulatory compliance and the protection of health, safety and fundamental rights. An ISO 42001 system scoped around business goals will need re-scoping, not just new documents.

Equally, EN 18286 is not a replacement for ISO 42001. It has no dedicated AI system impact assessment clause, and it covers only providers of high-risk systems. If you also deploy AI, build lower-risk systems, or want a certificate buyers recognise, ISO 42001 still earns its place. Around 350 organisations held ISO 42001 certification by spring 2026, and enterprise buyers increasingly ask for it.

Which one should you use?

Your situationWhat to do
You provide high-risk AI in the EUImplement EN 18286 for the presumption. Use ISO 42001 underneath if you want an organisation-wide AI governance system.
You already hold ISO 42001 and provide high-risk AIRun a gap analysis against EN 18286 and Article 17(1)(a)–(m), then add the regulatory layer.
You already hold ISO 9001 and provide high-risk AIEN 18286 also layers on ISO 9001. Add EN 18286, then consider ISO 42001 for AI-specific governance.
You only deploy AI built by othersEN 18286 isn't aimed at you. ISO 42001 is the better fit, alongside your deployer duties under Article 26.
Your AI isn't high-riskISO 42001 for governance and buyer assurance. EN 18286 is optional.
You are an SME or start-upCheck the Digital Omnibus simplifications: a simplified quality management route and simplified technical documentation are available.

The timeline

  • Dec 2023

    ISO/IEC 42001 published.

  • 1 Aug 2024

    EU AI Act enters into force.

  • Oct 2025

    prEN 18286 goes to public enquiry.

  • 12 Jul 2026

    CEN approves EN 18286.

  • 2 Aug 2026

    AI Act provisions on harmonised standards and conformity assessment apply.

  • 14 Oct 2026

    EN 18286:2026 cited in the Official Journal: the first AI Act harmonised standard.

  • 2 Dec 2027

    High-risk obligations apply to Annex III systems.

  • 2 Aug 2028

    High-risk obligations apply to AI in Annex I regulated products.

A 6-step plan for high-risk providers

  1. Confirm you are a provider of high-risk AI

    Check each system against Annex I and Annex III and the Article 6(3) exceptions. EN 18286's presumption only matters if Article 17 applies to you.

  2. Get the standard and read the citation

    Buy EN 18286:2026 from your national standards body and read the Official Journal citation for any limits on the presumption.

  3. Gap-assess against Article 17(1)(a)–(m)

    Use EN 18286's clauses as the scaffold. Mark what your ISO 42001 or ISO 9001 system already evidences, and what is missing.

  4. Build the regulatory layer

    Write a compliance strategy for each high-risk system, a technical documentation set aligned to Annex IV, serious-incident and authority communication procedures, and post-market monitoring plans.

  5. Cover the requirements EN 18286 doesn't

    Until more harmonised standards are cited, document how you meet Articles 9, 10 and 12–15 using drafts, state of the art and your own evidence.

  6. Keep ISO 42001 for the whole organisation

    Use it for AI you deploy, lower-risk systems, impact assessments and buyer assurance. Run one integrated management system, not two.

Common mistakes to avoid

  • Claiming an ISO 42001 certificate shows AI Act conformity
  • Assuming EN 18286 covers all high-risk requirements, rather than just the quality system
  • Treating a CEN-published standard as giving a presumption before Official Journal citation
  • Dropping ISO 42001 because EN 18286 exists
  • Keeping an ISO 42001 scope built around business goals, not health, safety and fundamental rights

A prompt to gap-assess your quality system

Paste in a summary of your current management system to get a first-draft Article 17 gap analysis. Have your compliance lead check it against the standard itself, which you need to buy.

Article 17 / EN 18286 gap analysis prompt
You are an EU AI Act compliance specialist for providers of
high-risk AI systems.

Our situation:
[high-risk AI systems we provide and their Annex III or Annex I
category; existing certifications (ISO 42001, ISO 9001, ISO 27001);
company size; whether any system keeps learning after release]

Our current management system:
[paste policies, procedures and records we already maintain]

Produce:
1. A table of the 13 elements of AI Act Article 17(1), (a) to (m),
   with columns: element, our existing evidence, status
   (covered / partial / missing), and the action needed.
2. Which existing ISO 42001 or ISO 9001 processes can be reused, and
   what must be added for regulatory purposes.
3. The documents we still need: compliance strategy per system,
   Annex IV technical documentation, serious-incident procedure,
   authority communication procedure, post-market monitoring plan.
4. High-risk requirements outside Article 17 (Articles 9, 10, 12-15)
   that our quality system does not yet evidence.
5. The ten highest-priority actions before 2 December 2027.

Do not quote the text of EN 18286. Refer to AI Act articles only.
Use only the facts I provide and mark anything uncertain UNKNOWN.
ISO 42001 shows you govern AI well. EN 18286 shows a regulator your quality system meets the law. If you provide high-risk AI in Europe, you want the second, and you'll build it faster on top of the first. — Kumail Mehdi

Frequently asked questions

Does ISO 42001 give a presumption of conformity under the EU AI Act?

No. ISO 42001 is not a European harmonised standard and is not cited in the Official Journal. It can support your compliance evidence, but it creates no legal presumption.

What does EN 18286 give a presumption of conformity for?

The quality management system required by Article 17(1) of the AI Act and part of the technical documentation obligation in Article 11(1), for providers of high-risk AI systems. It does not cover the other high-risk requirements such as risk management, data governance or human oversight.

When was EN 18286 cited in the Official Journal?

On 14 October 2026, through Commission Implementing Decision (EU) 2026/2274. It was the first harmonised standard cited under the AI Act.

Is EN 18286 mandatory?

No. Using harmonised standards is voluntary. Providers can meet Article 17 in other ways, but then they must demonstrate compliance without the benefit of the presumption.

Can I be certified to EN 18286?

EN 18286 is used in AI Act conformity assessment rather than as a separate voluntary certification like ISO 42001. Some certification bodies may offer assessments against it; check what any certificate actually attests.

If I'm ISO 42001 certified, how much work is EN 18286?

Less than starting from scratch. The management-system core carries over. The main additions are a regulatory compliance strategy, Annex IV technical documentation, serious-incident and authority communication procedures, post-market monitoring, and a scope focused on health, safety and fundamental rights.

Master AI governance in 6 months — 15 minutes a week

Join AI Governance – Decoded, my weekly newsletter that tracks AI Act standards, ISO 42001 and the frameworks around them in one short, practical read. And if you want to talk through your own path to presumption of conformity, connect with me on LinkedIn.

Subscribe to AI Governance – Decoded

Sources