ISO 42001 vs the EU AI Act: Which Governs Your AI?

ISO 42001 is a voluntary, certifiable AI management system standard. The EU AI Act is binding law. They are not interchangeable: certification does not make you legally compliant, and the Act does not require it. An ISO 42001 management system is a strong foundation for meeting the Act, not a substitute for it.

Two different instruments get treated as one. A vendor holds up an ISO 42001 certificate and calls it EU AI Act readiness. It is not. ISO 42001 is a management system you choose to build and certify. The EU AI Act is a law that binds you whether you certify anything or not. Knowing where the two meet, and where they do not, decides whether your AI governance spend actually reduces legal exposure.

Is ISO 42001 the same as complying with the EU AI Act?

No. One is voluntary, the other is binding. ISO/IEC 42001:2023 is the first certifiable AI management system standard, published in December 2023. An organisation adopts it by choice and an accredited body certifies it. The EU AI Act is a regulation that came into force on 1 August 2024 and applies on a phased timetable. Certification can show good governance, but it does not discharge a single legal obligation under the Act. Equally, the Act never asks for an ISO 42001 certificate. They answer different questions: one proves your management system works, the other tells you what the law demands of a given AI system.

What is the EU AI Act, and who does it bind?

The EU AI Act regulates AI by risk, not by organisation. It sorts systems into tiers: a small set of prohibited practices, a defined high risk category that carries the heavy duties (risk management, data governance, technical documentation, logging, human oversight, accuracy and robustness), limited risk systems that owe transparency, and minimal risk systems left largely alone. The obligations attach to the system and its risk class, not to a certificate on the wall.

Reach is the part Australian teams underestimate. The Act binds providers and deployers who place AI on the EU market, or whose system output is used in the EU, wherever the company itself sits. An Australian SaaS firm with EU customers can be in scope. Penalties run to EUR 35 million or 7 percent of worldwide annual turnover for the most serious breaches, with lower tiers beneath that.

The timetable is moving. Prohibited practices applied from 2 February 2025 and general purpose AI model obligations from 2 August 2025. The high risk obligations were set for 2 August 2026, but a provisional agreement reached on 7 May 2026 under the European Commission Digital Omnibus would defer stand alone high risk systems to 2 December 2027, and AI embedded in regulated products to 2 August 2028. That deferral is not yet law. It takes effect only once published in the Official Journal. Treat the dates as live, not fixed.

What does ISO 42001 actually give you?

A governance system, not a compliance certificate for any one law. ISO/IEC 42001 sets management system requirements across clauses 4 to 10, the same Annex SL backbone as ISO 27001, plus 38 Annex A controls across nine objectives selected through a Statement of Applicability. Its distinct demand is the AI system impact assessment: an assessment of the consequences of an AI system for individuals, groups and society, not only risk to the organisation. You certify it on a three year cycle. What you get is repeatable governance: roles, policies, risk and impact assessment, human oversight, monitoring and continual improvement. That scaffold is most of what a regulator, a customer or the EU AI Act will later ask you to evidence. For the detail, read what ISO 42001 is and why AI governance matters now.

Where do ISO 42001 and the EU AI Act meet, and where do they not?

ISO/IEC 42001EU AI Act
What it isVoluntary AI management system standardBinding EU regulation
Legal statusNot law; certified by choiceLaw; enforceable, with penalties
Unit of focusYour organisation management systemThe AI system and its risk class
How you show itAccredited certificate, three year cycleConformity with legal duties by risk tier
ReachGlobal, demand drivenExtraterritorial: EU market or EU used output
Cost of failureLost certificate, lost trustFines to EUR 35m or 7% of global turnover
StatusPublished December 2023, in use nowIn force August 2024, phased; high risk dates moving

The overlap is real but partial. The Act lets the Commission grant a presumption of conformity to organisations that follow harmonised standards it lists in the Official Journal. ISO 42001 is not one of those listed standards, so holding the certificate does not buy that legal presumption. A dedicated European AI management standard is being prepared to fill the gap, and it is being mapped against ISO 42001 Annex A controls so certified organisations can reuse the work. Until such a standard is listed, an ISO 42001 certificate is a credibility lift and a head start on evidence, not a legal shield.

Where they genuinely align is the governance machinery. The Act high risk duties (risk management, data governance, human oversight, logging, post market monitoring) map closely onto the AIMS, and the ISO 42001 impact assessment is close kin to the Act fundamental rights impact assessment. Build the management system once and most of the Act quality and risk obligations have somewhere to live.

Does an Australian organisation need either?

It depends on your market and your buyers, not your postcode. If your AI touches the EU market, or its output is used in the EU, the Act can bind you directly; that is a legal question to settle first. ISO 42001 is not mandated anywhere in Australia. Demand for it is commercial: customers, tenders and partners asking for proof your AI is governed. Locally, the Department of Industry, Science and Resources published a Voluntary AI Safety Standard in September 2024 with ten guardrails, and consulted on mandatory guardrails for high risk settings. Those guardrails align closely with ISO 42001, so the management system doubles as a practical route to meet them. The sensible order for most Australian organisations: work out whether the EU AI Act applies to you, then use an ISO 42001 management system as the governance backbone that serves the Act, the local guardrails and your customers at once. Cybernion builds ISO 42001 AI management systems to that end.

Frequently asked questions

Does ISO 42001 certification make me compliant with the EU AI Act?

No. ISO 42001 is a voluntary management system standard. The EU AI Act is binding law with obligations set by a system risk tier. Certification evidences good governance and overlaps with many of the Act duties, but it does not by itself satisfy the Act.

Does the EU AI Act require ISO 42001?

No. The Act does not name ISO 42001. Legal presumption of conformity comes only from harmonised standards the Commission lists in the Official Journal, and ISO 42001 is not currently one of them. A dedicated European standard is being prepared for that role.

When do the EU AI Act high risk rules apply?

Prohibited practices applied from February 2025 and general purpose AI obligations from August 2025. High risk obligations were set for August 2026, but a May 2026 provisional agreement under the Digital Omnibus would defer them to December 2027. That change is not yet formally adopted.

Does the EU AI Act apply to an Australian company?

It can. The Act is extraterritorial: it binds organisations that place AI on the EU market or whose system output is used in the EU, regardless of where the company is based. An Australian provider with EU customers should assess its exposure.


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 and ISO 42001 Lead Implementer, and speaks regularly at industry conferences.

Sources:

  1. ISO/IEC 42001:2023, the AI management system standard (ISO and IEC), December 2023
  2. EU AI Act, European Commission, 2024
  3. AI Act implementation timeline, European Commission AI Act Service Desk, 2026
  4. Voluntary AI Safety Standard, Department of Industry, Science and Resources, September 2024

Last updated: 21 June, 2026