In brief

The EU AI Act binds non-EU AI providers whose systems or outputs reach the EU. ISO 42001 does not replace it but supplies much of the governance evidence.

ISO 42001 vs EU AI Act is a comparison between a voluntary standard and a law. The EU AI Act is binding EU legislation that also applies to companies outside the EU, including Vietnamese software and AI vendors, when they place AI systems on the EU market or their AI output is used in the EU. ISO/IEC 42001 is a voluntary management system standard for AI. An ISO 42001 certificate does not prove AI Act compliance, but it builds much of the process and documentation the Act expects.

This post is for EU and international buyers assessing an AI vendor in Vietnam, and for Vietnamese AI and outsourcing companies deciding what to prepare before selling into the EU. For the standard itself, start with the ISO 42001 certification guide.

How do the EU AI Act and ISO 42001 differ?

The EU AI Act (Regulation (EU) 2024/1689) entered into force on 1 August 2024 and applies in stages. It sets obligations by role (provider, deployer, importer, distributor) and by the risk level of the AI system, backed by conformity assessment, registration and penalties.

ISO/IEC 42001:2023 (iso.org) is a management system standard. An organization builds its own AI governance processes, assesses AI risks and impacts, selects controls from Annex A, and can then be audited and certified by an independent certification body.

CriterionEU AI ActISO 42001
NatureEU law, mandatory within its scopeVoluntary standard
Who it coversAI systems and AI models placed on the EU market or whose output is used in the EUAny organization that develops, provides or uses AI, anywhere
How you demonstrate itConformity assessment, technical documentation, registration of high-risk systemsStage 1 and Stage 2 audit by a certification body
Consequence of failingAdministrative fines under Article 99No certificate or loss of it; possibly lost contracts

When does the EU AI Act apply to a company outside the EU?

Under Article 2, the Act applies to:

  • Providers placing AI systems or general-purpose AI models on the EU market, wherever they are established.
  • Deployers (organizations using AI) established or located in the EU.
  • Providers and deployers outside the EU where the output of the AI system is used in the EU.

A Vietnamese company does not need an EU office to be in scope. An AI product sold to EU customers, a model embedded in an EU company's product, or an AI service whose output is used in the EU can each be enough. Research and development before an AI system is placed on the market is excluded, as is purely personal, non-professional use. Non-EU providers of high-risk systems must appoint an authorised representative in the EU (Article 22); non-EU providers of general-purpose AI models have a similar duty (Article 54).

For outsourcing vendors, the role question matters most. A vendor that builds an AI feature to a client's specification, which the client then markets under its own name, is usually not the provider; the client is. A vendor that sells its own AI product or API into the EU usually is. Check the role per system, and write it into the contract.

Which risk levels does the EU AI Act define?

LevelExamplesMain obligations
Prohibited (Article 5)Social scoring, harmful manipulation of behaviour, exploiting vulnerabilities of vulnerable groupsMay not be placed on the market or used
High-risk (Annex III and Annex I)Recruitment and CV screening, credit scoring, education, biometrics, AI in medical devicesRisk management, data governance, technical documentation, logging, human oversight, conformity assessment, registration
Transparency obligations (Article 50)Chatbots, AI-generated content, deepfakesTell users they are interacting with AI; label generated content
Minimal riskSpam filters, AI in video gamesNo specific obligations beyond AI literacy (Article 4)

For most Annex III high-risk systems, the provider carries out the conformity assessment itself under the internal control procedure; biometric systems may need a notified body (Article 43). Providers must register high-risk systems in the EU database before placing them on the market (Article 49).

When do EU AI Act obligations apply?

The Digital Omnibus on AI (Regulation (EU) 2026/1744) moved the high-risk dates back. The dates below follow that amendment; check the consolidated text on EUR-Lex before relying on a specific date for a contract.

DateWhat applies
1 August 2024Regulation enters into force
2 February 2025Prohibited practices and AI literacy obligations
2 August 2025Obligations for general-purpose AI models
2 August 2026Transparency obligations (Article 50)
2 December 2027Stand-alone high-risk systems (Annex III), moved from 2 August 2026
2 August 2028High-risk AI in regulated products (Annex I)

The Omnibus also adds prohibited practices related to AI-generated non-consensual intimate imagery and child sexual abuse material, which the amending text schedules from 2 December 2026. Confirm the exact wording and date in the published regulation.

How far does ISO 42001 support EU AI Act compliance?

An AI management system (AIMS) under ISO 42001 produces processes and records that overlap with many obligations of high-risk system providers. Overlap is not automatic compliance: each obligation still has to be checked against the specific requirements of the relevant Article.

EU AI Act obligation (high-risk)Where ISO 42001 helps
Risk management system (Article 9)Clauses 6.1.2 to 6.1.4 and 8.2 to 8.4; A.5 assessing impacts of AI systems
Data and data governance (Article 10)A.7 Data for AI systems
Technical documentation (Article 11)A.6.2.7 AI system technical documentation
Record-keeping (Article 12)A.6.2.8 AI system recording of event logs
Transparency and information to deployers (Article 13)A.8.2 System documentation and information for users
Human oversight (Article 14)A.9 Use of AI systems (partial)
Quality management system (Article 17)Clauses 4 to 10

What ISO 42001 does not cover: registration in the EU database, the conformity assessment procedure and CE marking, appointing an EU representative, and the specific obligations for general-purpose AI models. For AI risk assessment method, ISO/IEC 23894 gives guidance on AI risk management that many teams use alongside ISO 42001. The AI inventory and risk records that feed this work are covered in AI system inventory and risk register for ISO 42001.

What are the obligations for general-purpose AI models?

Providers of general-purpose AI models (foundation models usable for many purposes) must draw up technical documentation, give information to downstream providers that integrate the model, have a policy to comply with EU copyright law and publish a summary of the training content. Models trained with more than 10^25 FLOPs are presumed to have systemic risk (Article 51) and carry extra duties: model evaluation, adversarial testing and incident reporting. Most Vietnamese startups use or fine-tune someone else's model, so they are usually providers of an AI system rather than of a general-purpose model; re-check the role before releasing a model to others.

What are the penalties under the EU AI Act?

Article 99 sets maximum fines by type of infringement: up to EUR 35 million or 7% of total worldwide annual turnover for the preceding year for prohibited practices; up to EUR 15 million or 3% for most other obligations; up to EUR 7.5 million or 1% for supplying incorrect information to authorities. For SMEs, including start-ups, the lower of the two amounts applies.

How should a Vietnamese AI company prepare, and what should EU buyers ask?

  1. Define scope: which AI systems have users, customers or output in the EU, and what your role is for each.
  2. Classify risk against Article 5, Annex III and Article 50; products touching recruitment, credit, education or health need a careful look.
  3. For high-risk systems, plan technical documentation, risk management and conformity assessment ahead of 2 December 2027 (Annex III) or 2 August 2028 (Annex I).
  4. Ensure AI literacy for staff working with AI, an obligation in effect since 2 February 2025.
  5. Use ISO 42001 as the governance frame for impact assessments, the AI system inventory and lifecycle records. The ISO 42001 implementation checklist lists the steps, and integrating ISO 27001 and ISO 42001 shows how to reuse an existing ISMS.
  6. Follow guidance from the European AI Office and the harmonised standards still being developed.

An EU buyer assessing a Vietnamese AI vendor can turn the same list into questions: which role does the vendor claim for this system, how has it classified the risk level, does it hold or plan ISO 42001 certification, and which AI Act obligations sit with the buyer as deployer rather than with the vendor.

How does pTrackly help?

pTrackly supports ISO 42001 as a framework: a compliance matrix by clause and Annex A, AI policy templates, recurring evidence tasks, a risk register, and read-only integrations with AI platforms to collect supporting evidence. pTrackly is not an EU AI Act compliance framework. It does not classify risk levels or perform conformity assessments for you; those decisions need your legal team or specialist advisers. See pTrackly for ISO 42001 and how long ISO 42001 certification takes.

Frequently asked questions

Does a Vietnamese company with no EU office have to comply with the EU AI Act?

Yes, if it places an AI system on the EU market or the system's output is used in the EU (Article 2). Non-EU providers of high-risk systems must also appoint an authorised representative in the EU.

Does ISO 42001 certification mean EU AI Act compliance?

No. ISO 42001 gives you risk management, documentation and logging processes, but it does not replace conformity assessment, registration or the Act's specific obligations.

When do high-risk obligations apply?

After the Digital Omnibus on AI, stand-alone high-risk systems under Annex III apply from 2 December 2027, and high-risk AI in regulated products under Annex I from 2 August 2028.

Is a chatbot a high-risk AI system?

Usually not. Chatbots fall under the Article 50 transparency obligations: users must be told they are interacting with AI. A chatbot becomes high-risk only if it is used for an Annex III purpose, such as screening job candidates.


References:

This article is for information only and is not legal advice.

Explore ISO 42001 with pTrackly
Tags:
ISO 42001EU AI ActAI GovernanceVietnamAI vendors