ISO 42001 vs NIST AI RMF: Which One a Canadian Company Needs

ISO 42001 vs NIST AI RMF for Canadian companies: what each one is, which buyers ask for, and how one set of AI evidence can answer both.

By Krikor Tengerian · Co-founder, SecuritAI Technologies Ltd. · October 2026

The question usually arrives in someone else’s form. An enterprise buyer’s security questionnaire now has an AI section, and it asks two things in a row: do you follow the NIST AI Risk Management Framework, and are you ISO 42001 certified? Your team has an AI feature in production and no clear answer to either. That is how most companies meet both frameworks, together, and it is not a sign you are behind.

Here is the part people get wrong

People treat them as rivals and assume they must pick one. They work at different levels.

The NIST AI RMF 1.0, published by the US National Institute of Standards and Technology on January 26, 2023 as NIST AI 100-1, describes itself as “intended to be voluntary”. Its core has four functions: Govern, Map, Measure and Manage. No one can certify you against it. NIST’s AI RMF page also says a revision is under way, so expect a new version.

ISO/IEC 42001:2023, published in December 2023, sets requirements for an AI management system: one you establish, maintain and keep improving. ISO’s own explainer says certification to it is voluntary and is done by independent certification bodies, not by ISO. In Canada, the Standards Council of Canada accredits the bodies that certify to ISO/IEC 42001.

The practical way to hold them: the AI RMF describes what good AI risk management looks like. ISO 42001 asks you to prove you run it, to an outside auditor, on a recurring cycle.

ISO 42001 vs NIST AI RMF, side by side

For the full detail of each one, read our guides to ISO 42001 in Canada and the NIST AI RMF for Canadian companies. This page is only about how they fit together.

Where Canada sits today

Canada has no AI specific federal law in force. The Artificial Intelligence and Data Act was part of Bill C-27, which never completed committee stage before that Parliament ended in January 2025, and the government’s AIDA page is now archived. The current Parliament’s Bill C-36, introduced June 15, 2026, is a privacy bill. It would require an explanation, on request, when an automated decision system makes a prediction, recommendation or decision that could have a legal or similarly significant effect on a person. It is at second reading and is not law.

What applies today is the law you already have, PIPEDA first, plus your sector and province:

• Quebec: section 12.1 of the private sector privacy act, added by Law 25, requires a business to inform a person when a decision about them is based exclusively on automated processing, and to explain the main factors on request.
• Ontario public sector: the Enhancing Digital Security and Trust Act, 2024 sets AI duties for public sector entities named by regulation. The AI sections are in force, but as of October 2026 no regulation names any entity, so the duties do not yet apply to anyone. Private companies are not covered.
• Federal institutions: the Treasury Board Directive on Automated Decision-Making applies to federal institutions subject to the Policy on Service and Digital, not to private companies.
• Voluntary: ISED’s Voluntary Code of Conduct on advanced generative AI systems, from September 2023, is a commitment signatories make. It does not change legal duties such as PIPEDA.

Neither ISO 42001 nor the AI RMF is required by Canadian law. Buyers require them. That is why the question arrives in a questionnaire and not in a letter from a regulator.

Which one first

Start with the AI RMF if your buyers ask whether you follow a recognized framework and nobody has asked for a certificate. It is free to read, and it gives your team a shared vocabulary for AI risk.

Plan for ISO 42001 if a contract, a tender or a European customer asks for third party assurance, or if you already hold ISO 27001 and run a management system.

Either way, build the evidence once. NIST’s AI Resource Center lists a crosswalk from the AI RMF to ISO/IEC 42001. A third party wrote it, and NIST notes that listing it does not mean NIST endorses it. Use it as a map, not as an authority.

One evidence set for both

Score each line implemented, partial or not started, and name the record you would hand an auditor or a buyer. The mapping is our own reading of both documents.

1. AI system inventory: every model, vendor API and AI feature in use, with its owner and purpose. AI RMF: Govern and Map. ISO 42001: the scope of the management system. Our guide covers what to record for every AI system.
2. AI policy and accountable roles: a signed policy and named owners. AI RMF: Govern. ISO 42001: leadership, policy and roles.
3. Risk and impact assessment per system: who could be harmed, how, and what you decided. AI RMF: Map and Measure. ISO 42001: AI risk assessment and impact assessment.
4. Data governance: where training and prompt data come from, where they are stored, and for how long. AI RMF: Map and Manage. ISO 42001: data controls. In Canada this is also your PIPEDA record.
5. Testing before launch and after changes: adversarial test results, including prompt injection and data leakage, with dates. AI RMF: Measure. ISO 42001: verification and validation. This is the line most teams cannot fill, and it is the one AI red teaming produces.
6. Monitoring and incident records: what you watch in production, what raised an alert, and what you did about it. AI RMF: Manage. ISO 42001: operation, monitoring and corrective action.
7. Third party AI: agreements and assessments for every AI vendor you rely on. AI RMF: Govern and Manage. ISO 42001: supplier controls.
8. Internal audit and management review: ISO 42001 requires both. The AI RMF does not. This line separates following a framework from being ready for certification.

The gap is almost never conceptual

Teams usually understand the risks of their own AI better than either framework does. Where they get stuck is evidence. They cannot show a buyer the inventory, the test results from before launch, or the record of who approved the model going live. Both frameworks reward the same thing: a record made at the time, not a policy written the week before the audit.

Being straight about our role

We are not an auditor and we are not a certification body. We cannot certify you to ISO 42001, and nobody can certify you to the NIST AI RMF. What SecuritAI does is produce the testing and runtime evidence in lines 5 and 6: adversarial testing before launch and a firewall that logs what it blocks in production. For the controls and documents behind the other lines, our compliance software SecuritComply tracks ISO 42001 and the NIST AI RMF next to PIPEDA, SOC 2 and ISO 27001.

References

All links tested on October 23, 2026. The iso.org, scc-ccn.ca and legisquebec.gouv.qc.ca pages return 403 to automated fetchers and were read directly in a browser.

1. ISO, ISO/IEC 42001:2023, AI management systems
2. ISO, ISO/IEC 42001 explained, section on certification
3. Standards Council of Canada, Artificial intelligence management systems accreditation
4. NIST, Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1
5. NIST, AI Risk Management Framework overview
6. NIST AI Resource Center, AI RMF Core
7. NIST, Generative AI Profile, NIST AI 600-1
8. NIST AI Resource Center, Crosswalks
9. Parliament of Canada, LEGISinfo, Bill C-27 (44-1) and Bill C-36 (45-1)
10. Innovation, Science and Economic Development Canada, Artificial Intelligence and Data Act (archived)
11. Innovation, Science and Economic Development Canada, Voluntary Code of Conduct on the Responsible Development and Management of Advanced Generative AI Systems
12. Treasury Board of Canada Secretariat, Directive on Automated Decision-Making
13. LegisQuébec, Act respecting the protection of personal information in the private sector, section 12.1
14. Government of Ontario, e-Laws, Enhancing Digital Security and Trust Act, 2024

← Back to the blog