Compliance Colleague · Guides

EU AI Act vs ISO/IEC 42001: what the standard covers, and what it can't

By Edward Casteloes, Founder of Compliance Colleague · 14+ years in GRC · November 10, 2026

The short version: ISO/IEC 42001:2023 gives you the management system the EU AI Act assumes you have, and most of the Act’s high-risk obligations have a 42001 clause or control to hang from. What 42001 cannot give you is legal effect. A 42001 certificate carries no presumption of conformity, is not a conformity assessment, and does not produce the registration, the declaration of conformity, the CE marking, the Annex IV documentation or the incident reports the Act asks for. The one standard that does carry a presumption, for the quality management system and part of the technical documentation, is EN 18286:2026, cited in the Official Journal on 14 October 2026.

The table maps the obligations of Regulation (EU) 2024/1689, as amended by Regulation (EU) 2026/1744 (the Digital Omnibus on AI, in force 27 July 2026), to ISO/IEC 42001:2023 clauses and Annex A controls. Provider obligations for high-risk AI systems come first, then the duties of deployers and the deliverables that have no equivalent in any management system. It’s our own crosswalk, not an official ISO or Commission mapping.

EU AI ActISO/IEC 42001:2023CoverageNotes
Art. 9
Risk management system (continuous, iterative, across the life cycle)
Clause 6.1.2 AI risk assessment; 6.1.3 AI risk treatment
Clause 8.2 and 8.3 (performing the assessment and treatment)
A.5.2 AI system impact assessment process
A.6.2.4 AI system verification and validation
Broad42001's risk assessment and treatment cycle is the right shape. Art. 9 is narrower in one way and wider in another: it is about risks to health, safety and fundamental rights from the system's intended use and reasonably foreseeable misuse, and it requires testing against defined metrics before placing on the market. Keep the register, add the Art. 9 lens and the test evidence.
Art. 10
Data and data governance (training, validation and test data)
A.7.2 Data for development and enhancement
A.7.3 Acquisition of data; A.7.4 Quality of data
A.7.5 Data provenance; A.7.6 Data preparation
A.4.3 Data resources
BroadThe A.7 controls cover the same ground: provenance, quality, preparation, bias examination. Art. 10 adds specifics 42001 doesn't state, including the relevance and representativeness of data for the intended purpose, examination for biases likely to affect health, safety or fundamental rights, and the conditions in the new Art. 4a for using special-category personal data to detect and correct bias.
Art. 11
Technical documentation (Annex IV content, kept up to date)
A.6.2.7 AI system technical documentation
A.6.2.3 Documentation of design and development
A.6.2.2 AI system requirements and specification
Clause 7.5 Documented information
Partial42001 asks for technical documentation; the Act dictates its contents (Annex IV) and expects it to show conformity with Section 2 before placing on the market. SMEs, start-ups and small mid-caps may use the Commission's simplified form (Art. 11(1) as amended). EN 18286 now carries a presumption for the first sentence of Art. 11(1) when applied; 42001 does not.
Art. 12
Record-keeping: automatic logging over the system's lifetime
A.6.2.8 AI system recording of event logs
PartialA.6.2.8 asks you to decide at which life-cycle phases logging is enabled. Art. 12 requires logging to be designed in, covering at least the period of use, the reference database checked, the matching input data and the identity of the people verifying results for biometric systems. Retention is a separate duty: providers keep logs under their control for at least six months (Art. 19); so do deployers (Art. 26(6)).
Art. 13
Transparency and instructions for use to deployers
A.8.2 System documentation and information for users
A.10.4 Customers
A.9.4 Intended use of the AI system
Partial42001 says inform users proportionately. Art. 13 prescribes the instructions for use: provider identity, intended purpose, accuracy level and metrics, known and foreseeable risks, human oversight measures, expected lifetime and maintenance. Treat Art. 13(3) as the checklist and A.8.2 as the process that produces it.
Art. 14
Human oversight (designed in; effective; stop and override)
A.9.2 Processes for responsible use of AI systems
A.6.1.3 Processes for responsible AI system design and development
A.5.4 Assessing impact on individuals or groups
Partial42001 expects human oversight as part of responsible use and design, without saying what it must do. Art. 14 does: the overseer must be able to understand the system's capacities and limits, stay aware of automation bias, interpret output, decide not to use it, and stop it. For remote biometric identification, two people must verify a match before action (Art. 14(5)).
Art. 15
Accuracy, robustness and cybersecurity
A.6.2.4 Verification and validation
A.6.2.6 Operation and monitoring
Clause 9.1 Monitoring, measurement, analysis and evaluation
Partial42001 verifies and monitors; the Act asks for an appropriate level of accuracy, robustness and cybersecurity, with accuracy levels and metrics declared in the instructions for use. Neither sets a numeric threshold. Cybersecurity itself is outside 42001 (A.4.5 and A.6.2.8 are documentation controls); it comes from your ISMS. A high-risk system within the Cyber Resilience Act that meets its Art. 12(1) conditions is deemed to meet Art. 15's cybersecurity requirements (Art. 42(3) as amended).
Art. 17
Quality management system (13 required components)
Clauses 4 to 10 (the management system itself)
A.2.2 AI policy; A.3.2 AI roles and responsibilities
A.6.1.2 and A.6.1.3 (objectives and processes for development)
A.10.3 Suppliers; A.8.4 Communication of incidents
PartialAn AIMS is a management system, so clauses 4 to 10 host most of Art. 17(1): policy, roles, resources, document control, internal audit, management review, improvement. The Act's list adds items 42001 doesn't name: a regulatory-compliance strategy including conformity assessment and the handling of modifications, a post-market monitoring system (Art. 72) and serious-incident reporting (Art. 73), and communication with competent authorities. EN 18286 carries the presumption for Art. 17(1); 42001 does not.
Art. 26
Deployer obligations (use per instructions, oversight, logs, workers, affected persons)
A.9.2 Processes for responsible use; A.9.3 Objectives for responsible use
A.9.4 Intended use of the AI system
A.8.5 Information for interested parties
A.6.2.8 Event logs; A.6.2.6 Operation and monitoring
PartialThe A.9 controls are written for organisations that use AI, so a deploying organisation with 42001 has the process. The Act adds specifics: human oversight by people with the competence, training and authority to act (26(2)); input data relevance where the deployer controls it (26(4)); logs kept at least six months (26(6)); informing workers' representatives before workplace use (26(7)); informing people subject to decisions (26(11)). Don't read the ISO 'AI deployer' label as the Act's 'deployer': they are different definitions.
Art. 27
Fundamental rights impact assessment (FRIA) for certain deployers
Clause 6.1.4 AI system impact assessment (AISIA)
A.5.2 to A.5.5 (impact assessment process, documentation, individuals, society)
PartialThe AISIA is the natural input, and the Act lets deployers reuse an existing GDPR DPIA by cross-reference (27(4) as amended). But the FRIA is a separate legal deliverable with fixed content (27(1)(a) to (f)) and a notification to the market surveillance authority, and it is owed only by public bodies, private entities providing public services, and deployers of credit-scoring and life and health insurance pricing systems. A private employer screening CVs owes no FRIA; it still owes Art. 26.
Arts. 49, 71
Registration in the EU database before placing on the market
No equivalentNoneNo 42001 equivalent. Providers of Annex III high-risk systems register the system (and themselves) in the EU database; a provider that concludes its Annex III system is not high-risk under Art. 6(3) still registers that conclusion (Art. 49(2)). Deployers that are public authorities register their use.
Arts. 43, 47, 48
Conformity assessment, EU declaration of conformity, CE marking
No equivalentNoneA certificate of a management system is not a conformity assessment of an AI system. For Annex III points 2 to 8 the route is internal control (Annex VI); for point 1 (biometrics) a notified body is mandatory unless harmonised standards or common specifications were applied to all the requirements. Then the declaration (Art. 47) and the CE marking (Art. 48). 42001 changes none of this.
Arts. 72, 73
Post-market monitoring and serious-incident reporting
A.6.2.6 Operation and monitoring
A.8.4 Communication of incidents; A.8.3 External reporting
Clause 10 Improvement
Partial42001 monitors and communicates incidents in general terms. The Act requires a documented post-market monitoring plan (Commission guidance and template due by 2 September 2027 under the Omnibus) and reporting of serious incidents to the market surveillance authority within the Act's deadlines: immediately after establishing the causal link and no later than 15 days, 10 days for a death, and 2 days for a widespread infringement or a critical-infrastructure disruption.

Annex A numbers follow ISO/IEC 42001:2023: A.2 policies, A.3 internal organisation, A.4 resources, A.5 impact assessment, A.6 life cycle, A.7 data, A.8 information for interested parties, A.9 use, A.10 third parties. Check each against your Statement of Applicability; several of the A.4, A.6 and A.7 controls apply to providers only.

Why a 42001 certificate gives no presumption of conformity

Article 40(1) presumes conformity only for high-risk AI systems that conform to harmonised standards whose references are published in the Official Journal. A harmonised standard is a European standard adopted on a Commission request (Regulation (EU) No 1025/2012, Article 2(1)(c)). ISO/IEC 42001 is an international standard. It is not cited, and conformity with it presumes nothing.

The first harmonised standard is EN 18286:2026, Artificial intelligence – Quality management system for EU AI Act regulatory purposes. Commission Implementing Decision (EU) 2026/2274 published its reference in the Official Journal on 14 October 2026. Applied in full it gives a presumption for Article 17(1) and the first sentence of Article 11(1), and only for those. EN 18286 follows the same high-level structure as 42001 and its Annex C pairs its clauses with ISO/IEC 42001, without equating them: EN 18286 has no AI system impact assessment and no continual-improvement clause, and 42001 has both. In practice a 42001 AIMS is a sound place to host an EN 18286 QMS; it is not a substitute for one.

Dates after the Omnibus

ObligationApplies fromNotes
Prohibited practices (Art. 5)2 February 2025Already in force. The two new prohibitions added by the Omnibus (non-consensual intimate imagery and child sexual abuse material, Art. 5(1)(ba) and (bb)) apply from 2 December 2026.
GPAI model obligations (Chapter V)2 August 2025Fines under Art. 101 from 2 August 2026. Models placed on the market before 2 August 2025 have until 2 August 2027.
Transparency duties (Art. 50)2 August 2026Generative systems placed on the market before that date must meet the machine-readable marking duty in Art. 50(2) by 2 December 2026.
High-risk systems, Annex III (Art. 6(2))2 December 2027Was 2 August 2026 before Regulation (EU) 2026/1744 (the Digital Omnibus on AI), in force 27 July 2026.
High-risk systems, Annex I products (Art. 6(1))2 August 2028Was 2 August 2027. Systems already on the market are caught only if their design is significantly changed on or after the date (Art. 111(2) as amended).
Legacy high-risk systems used by public authorities2 August 2030Unchanged by the Omnibus.

What you can reuse

How to use this

  1. Settle your role under the Act first. Provider, deployer, or both (a deployer that rebrands, substantially modifies or repurposes a system becomes the provider under Article 25(1)). The Act’s definitions, not the ISO role labels, decide which rows apply.
  2. Classify each system. Prohibited, high-risk (Annex I or Annex III), limited-risk transparency duties, or none of these. Only high-risk systems carry the Article 9 to 17 obligations in the table.
  3. Run a gap check on the “Partial” and “None” rows against the dates above, and decide whether EN 18286 is worth applying for the Article 17 presumption.

If your AIMS sits on an existing ISMS, see how ISO 27001’s clauses and Annex A fit together: 42001 shares the clause 4 to 10 structure, and your ISMS supplies the cybersecurity Article 15 expects.

Frequently asked questions

Does ISO 42001 certification mean we comply with the EU AI Act?

No. ISO/IEC 42001 is an international standard, and the AI Act's presumption of conformity (Article 40) applies only to European harmonised standards whose references are published in the Official Journal. ISO 42001 is not one of them. A 42001 certificate shows you run an AI management system; it is not a conformity assessment of any AI system, and it gives no presumption for any article of the Act.

Which standard does give a presumption of conformity under the AI Act?

EN 18286:2026, Artificial intelligence, Quality management system for EU AI Act regulatory purposes. Its reference was published in the Official Journal on 14 October 2026 by Commission Implementing Decision (EU) 2026/2274. Applied in full, it gives a presumption of conformity for Article 17(1) (the quality management system) and the first sentence of Article 11(1) (technical documentation). Further harmonised standards for risk management, data, logging, transparency, human oversight, accuracy, robustness and cybersecurity are still in preparation under standardisation request C(2025)3871.

What does ISO 42001 cover that the AI Act does not require?

The management-system discipline: context and scope (clause 4), leadership (5), an AI policy (A.2), planned objectives (6.2), competence and awareness (7), internal audit and management review (9) and continual improvement (10). It also asks every organisation in scope, deployers included, to assess the impact of its AI systems on individuals and society (6.1.4, A.5), where the Act requires a fundamental rights impact assessment only from certain deployers of certain high-risk systems.

What does the AI Act require that ISO 42001 does not cover?

The legal deliverables: registration in the EU database before placing a high-risk system on the market (Article 49), a conformity assessment, an EU declaration of conformity and the CE marking (Articles 43, 47 and 48), technical documentation with the Annex IV content (Article 11), automatic logging with six-month retention (Articles 12, 19 and 26(6)), instructions for use with the Article 13(3) content, a post-market monitoring plan (Article 72) and serious-incident reports to the market surveillance authority within 15, 10 or 2 days (Article 73), and the deployer duties in Article 26, including informing workers' representatives before workplace use.

When do the high-risk obligations apply?

From 2 December 2027 for Annex III systems (biometrics, critical infrastructure, education, employment, essential services, law enforcement, migration, justice) and from 2 August 2028 for AI systems that are safety components of Annex I products. Both dates were moved by Regulation (EU) 2026/1744, in force 27 July 2026; they were 2 August 2026 and 2 August 2027 before it. The prohibitions, GPAI model obligations and Article 50 transparency duties already apply.

Is a 42001 AI system impact assessment the same as a FRIA?

No. The 42001 impact assessment (clause 6.1.4) is required for every AI system in the management system's scope and has no fixed content. The fundamental rights impact assessment (Article 27) is owed only by public bodies, private entities providing public services, and deployers of credit-scoring and life and health insurance pricing systems; it has fixed content, and its results are notified to the market surveillance authority. The 42001 assessment is good input, and since the Omnibus a FRIA may cross-reference an existing GDPR data protection impact assessment, but the FRIA is still its own deliverable.

Should we get ISO 42001 certified before the AI Act applies?

It depends on what you need to show. For a provider of a high-risk system, EN 18286 is now the standard that carries legal weight for the quality management system, and a 42001 AIMS can host it. For a deployer, or for an organisation whose AI is not high-risk, 42001 certification is the recognised way to show customers and regulators that AI is governed, and nothing in the Act requires it. Build the management system first; decide on certification once you know which role and which obligations you actually have.

Compliance Colleague answers questions like “Which EU AI Act obligations does our ISO 42001 Statement of Applicability leave uncovered, and what does Article 26 require of us as a deployer?” with the article, clause or control cited. Try it free, no card required →

ISO/IEC 42001 is copyrighted by ISO and IEC. This guide cites clause and control numbers and short titles only and paraphrases everything else. The mapping is our own research aid, not an official crosswalk. Confirm against the Official Journal text and the standard, and with your conformity assessment body or counsel, before relying on it. Not legal advice. Drafted with AI (Compliance Colleague), approved for publication by Edward Casteloes CISSP, CISA.