ISO 42001·11 min read

ISO 42001 vs EU AI Act: Overlap, Gaps and 2027 Deadlines

ISO 42001 vs EU AI Act comes down to one distinction: ISO/IEC 42001 is a voluntary standard for running an AI management system, and the EU AI Act is a binding law that sets duties by risk level. A certified ISO 42001 program covers much of the governance work the Act expects, but a certificate is not legal compliance.

This guide compares the two side by side, shows where the work overlaps, and lists the obligations ISO 42001 does not touch. It is built on the AI Act timeline as amended by the Digital Omnibus on AI, verified on 9 November 2026. For an introduction to the standard, start with the ISO 42001 guide.


The Short Answer

  • ISO 42001 tells you how to govern AI as an organization: policy, roles, risk and impact assessment, controls, internal audit, and continual improvement. You choose to adopt it, and an accredited body certifies it.
  • The EU AI Act (Regulation (EU) 2024/1689) tells you what you must do for specific AI systems based on their risk. It applies whether you like it or not, and national authorities enforce it with fines.

Think of ISO 42001 as the operating system and the AI Act as one of the applications that runs on it. A good AI management system makes AI Act compliance much easier to prove. It does not replace the legal steps the Act requires for high-risk systems.


What ISO 42001 Is

ISO/IEC 42001:2023 is the first international management system standard for artificial intelligence. It follows the same structure as ISO 27001 and ISO 9001, so teams that run those systems will recognize it.

Its core requirements are:

  • Context and scope. Decide which AI systems and roles are in scope, such as developer, provider, or user.
  • Leadership and AI policy. Top management owns AI governance and sets objectives.
  • AI risk assessment and treatment. Identify and treat risks to the organization.
  • AI system impact assessment. Assess how each AI system could affect people and groups.
  • Annex A controls. Select controls across areas such as data for AI systems, the AI system lifecycle, information for interested parties, and third-party relationships.
  • Internal audit and management review. Check the system works and improve it.

Certification runs on a three-year cycle with annual surveillance audits. The ISO 42001 certification post covers the path step by step.


What the EU AI Act Is

The EU AI Act is the first broad AI law in a major market. It entered into force on 1 August 2024 and applies in phases. It sorts AI into tiers by risk:

  • Prohibited practices. Uses banned outright, such as social scoring and certain manipulative or exploitative systems.
  • High-risk AI. Systems in sensitive areas listed in Annex III, such as hiring, credit scoring, education, and critical infrastructure, plus AI used as a safety component in products covered by Annex I.
  • Transparency obligations. Article 50 duties for systems that interact with people or generate content, such as chatbots and deepfakes.
  • General-purpose AI (GPAI) models. Duties for providers of foundation models.
  • Minimal risk. Everything else, with no specific duties beyond AI literacy.

The Act also assigns duties by role. A provider develops an AI system and places it on the market. A deployer uses one under its own authority. Most obligations for high-risk AI fall on providers.


EU AI Act Timeline After the Digital Omnibus

This is the time-sensitive part. The Digital Omnibus on AI changed the high-risk dates. According to the European Commission's AI Act page, checked on 9 November 2026, the Omnibus entered into force on 27 July 2026.

ObligationApplies from
Prohibited practices and AI literacy2 February 2025
General-purpose AI model rules and governance2 August 2025
General application, including Article 50 transparency rules2 August 2026
High-risk AI in Annex III use cases (employment, credit, education, biometrics, critical infrastructure)2 December 2027 (was 2 August 2026)
High-risk AI in Annex I regulated products (medical devices, machinery, toys)2 August 2028

Two practical points follow:

  • Transparency did not move. If your product has a chatbot or generates synthetic content, the Article 50 duties already apply.
  • The high-risk delay is a runway, not a reprieve. Building a quality management system, technical documentation, and post-market monitoring takes 12 to 18 months for most teams. December 2027 is closer than it looks.

Check the Commission's AI Act page before relying on these dates. Further amendments are possible.


ISO 42001 vs EU AI Act: Side by Side

ISO 42001EU AI Act
TypeVoluntary international standardBinding EU regulation
ScopeYour organization's AI management systemSpecific AI systems, by risk tier
Who it applies toAnyone who chooses to adopt itProviders, deployers, importers, distributors with an EU link
ProofCertificate from an accredited certification bodyConformity assessment, EU declaration of conformity, CE marking (high-risk)
EnforcementLoss of certificateFines under Article 99 of up to €35 million or 7% of worldwide annual turnover for prohibited practices
GeographyGlobalEU market, including non-EU companies selling into it
UpdatesRevised by ISO on a multi-year cycleAmended by EU legislation, such as the Omnibus

Where ISO 42001 Covers the AI Act

The overlap is real and large for high-risk providers. Here is how the main Act obligations map to an ISO 42001 system:

  • Risk management (Article 9). ISO 42001's risk assessment, risk treatment, and impact assessment cover most of the process. The Act wants risks assessed across the full lifecycle, including foreseeable misuse. Make sure your method says so.
  • Data and data governance (Article 10). Annex A controls on data for AI systems cover data quality, provenance, and preparation. The Act adds specific checks for bias and representativeness.
  • Technical documentation (Article 11). ISO 42001 requires lifecycle documentation. The Act prescribes the exact contents in Annex IV, so expect to reformat.
  • Record-keeping and logging (Article 12). The lifecycle controls cover event logging. The Act requires automatic logs built into the system.
  • Information for deployers (Article 13). Annex A controls on information for interested parties map well to the Act's instructions for use.
  • Human oversight (Article 14). ISO 42001 expects you to define how humans oversee AI. The Act is more prescriptive about design features that let people intervene.
  • Quality management system (Article 17). This is the strongest match. An ISO 42001 management system already has policy, roles, document control, internal audit, and corrective action.
  • Post-market monitoring (Article 72). The operation and monitoring controls give you the base process.
  • AI literacy (Article 4). ISO 42001's competence and awareness clauses cover training.

For a deployer, ISO 42001's impact assessment is also a solid base for a fundamental rights impact assessment (FRIA) under Article 27, where one is required. The method is close to a GDPR data protection impact assessment.


Where ISO 42001 Falls Short of the AI Act

Several AI Act duties have no ISO 42001 equivalent. These are legal and product steps, not management system steps:

  • Conformity assessment (Article 43). High-risk systems need a formal assessment before market entry. For most Annex III systems this is internal control, but it must follow the Act's procedure.
  • EU declaration of conformity and CE marking (Articles 47 and 48). No certificate replaces these.
  • Registration (Article 49). Many high-risk systems must be registered in the EU database.
  • Serious incident reporting (Article 73). Providers must report serious incidents to market surveillance authorities within set deadlines. ISO 42001 handles incidents internally, not to regulators.
  • Authorised representative (Article 22). Non-EU providers of high-risk systems must appoint one in the EU.
  • Accuracy, robustness, and cybersecurity (Article 15). ISO 42001 asks for verification and validation, but the Act sets performance expectations. ISO 27001 helps with the cybersecurity part.
  • Prohibited practice screening (Article 5). The Act bans some uses outright. Your AI policy should screen for them, but the standard does not list them.
  • GPAI model duties (Articles 53 to 55). Foundation model providers face documentation, copyright, and systemic risk duties that sit outside ISO 42001.

Harmonised Standards: Where EN 18286 Fits

Under the AI Act, following a harmonised standard cited in the EU Official Journal gives a presumption of conformity with the parts of the Act it covers. ISO 42001 is not one of them.

The first harmonised standard for the Act is EN 18286:2026, a quality management system standard for AI providers. CEN-CENELEC reported it as cited in the Official Journal in October 2026. It targets the Article 17 quality management system duty only. According to CEN-CENELEC's JTC 21 committee, it includes informative mappings to ISO/IEC 42001 and ISO 9001, so companies certified to either can extend their existing system.

Standards for the other high-risk requirements, such as risk management, data governance, logging, human oversight, and cybersecurity, are still in development. Until they are cited, you show conformity directly against the Act's text.

The practical takeaway: an ISO 42001 system plus an EN 18286 gap assessment is the most efficient route to an Article 17 quality management system.


Which Should You Start With?

Your answer depends on whether any of your AI is high-risk.

Your AI is high-risk under Annex III. You need AI Act compliance by 2 December 2027. Start with an AI system inventory and classification, then build an ISO 42001 management system as the backbone. Layer the Act-specific steps on top: Annex IV technical documentation, conformity assessment, registration, and incident reporting.

Your AI is not high-risk. Most SaaS AI features, such as summarization, search, and coding assistants, fall here. Your AI Act duties are mostly transparency and AI literacy. ISO 42001 is still worth it if enterprise buyers are asking how you govern AI, because a certificate answers that in one document.

You are unsure. Do the classification first. It is a few weeks of work and it decides everything else. Watch for Annex III areas like HR screening, credit decisions, or education scoring, which sometimes hide inside general SaaS products.

If you also use the NIST AI Risk Management Framework, the ISO 42001 vs NIST AI RMF comparison shows how all three fit together.


A Practical Path for AI SaaS Companies

  1. Inventory your AI systems. List every model and AI feature, including third-party APIs.
  2. Classify each one under the AI Act. Prohibited, high-risk, transparency, or minimal. Record your role as provider or deployer.
  3. Meet the duties already in force. Article 50 transparency notices apply now. So does Article 4 AI literacy, which the Omnibus rewrote as a duty to take measures that support staff literacy rather than guarantee a set level.
  4. Build the ISO 42001 management system. Policy, roles, risk and impact assessment, and Annex A controls.
  5. Add the Act-specific layer for high-risk systems. Technical documentation, logging, human oversight design, conformity assessment, registration, and incident reporting.
  6. Certify to ISO 42001. Use the certificate for buyers, and keep the AI Act file for regulators.
  7. Track the timeline. Recheck dates and new harmonised standards each quarter.

If you already run ISO 27001, reuse its document control, internal audit, and risk method. See compliance controls overlap for where the work double-counts.


For the full engagement, see the ISO 42001 service page. For background on the standard, read the ISO 42001 guide, and for audit steps, see ISO 42001 certification. To compare it with the US framework, read ISO 42001 vs NIST AI RMF.


Frequently Asked Questions

Does ISO 42001 certification mean I comply with the EU AI Act? No. A certificate shows your AI management system works. It is not a conformity assessment, it does not give you a CE marking, and ISO 42001 is not a harmonised standard. It does give you much of the governance and documentation evidence the Act asks for.

When do the EU AI Act high-risk rules apply? After the Digital Omnibus entered into force on 27 July 2026, Annex III high-risk rules apply from 2 December 2027 and Annex I product rules from 2 August 2028. Prohibited practices have applied since 2 February 2025.

Is ISO 42001 a harmonised standard under the AI Act? No. The first harmonised standard is EN 18286:2026, a quality management system standard cited in the Official Journal in October 2026. It maps to ISO 42001, so an existing system is a strong starting point.

Does the EU AI Act apply to US companies? Yes, if you place an AI system on the EU market or its output is used in the EU. Non-EU providers of high-risk systems must also appoint an EU authorised representative.

Should I get ISO 42001 if my AI is not high-risk? Often, yes, for commercial reasons. Buyers ask how you govern AI, and a certificate answers that. It also gives you a structure that scales if a future product becomes high-risk.


Ready to Map ISO 42001 to the EU AI Act?

ShieldKey Solutions helps AI SaaS companies classify their systems under the AI Act, build an ISO 42001 management system, and close the gap to high-risk obligations before December 2027. One program, built once, that serves both buyers and regulators.

Schedule a scoping call →