In this article7 sections
  1. What Article 4 actually requires
  2. Who has to comply
  3. What good measures look like in practice
  4. The five things worth having on file
  5. What enforcement actually looks like
  6. Practical first step
  7. References

Article 4 of the EU AI Act has been legally binding since 2 February 2025. National market surveillance authorities supervise it from 2 August 2026. And on 27 July 2026, the Digital Omnibus on AI (Regulation (EU) 2026/1744) rewrote it. There is no new deadline to race against: the obligation already applies. What matters is whether you can show the measures you have taken.

Here is what the regulation actually says now, who it applies to, and the five things worth having on file.

What Article 4 actually requires

The text of Article 4(1), as amended by Regulation (EU) 2026/1744, is short:

Providers and deployers of AI systems shall take measures to support the development of AI literacy of their staff and other persons dealing with the operation and use of AI systems on their behalf, taking into account their technical knowledge, experience, education and training and the context the AI systems are to be used in, and considering the persons or groups of persons on whom the AI systems are to be used. This obligation does not require providers or deployers to guarantee any specific level of AI literacy of any individual.

Four things matter in that text:

  1. It applies to "deployers", not only AI vendors. If your sales team uses ChatGPT, your developers use Copilot, or your CRM has built-in AI features, you are a "deployer" under the Act.

  2. It is an obligation of effort. You must take measures that support AI literacy. You do not have to guarantee a particular level for each person. The old wording ("ensure, to their best extent, a sufficient level") is gone.

  3. It is still binding. "Shall take measures" is a legal duty on the company. Doing nothing is not an option the new text allows.

  4. It is context-dependent. The training a finance analyst running an AI fraud model needs is not the same as what a marketing assistant using ChatGPT needs.

The Omnibus also added Article 4(2): the Commission and the Member States must support companies, in particular SMEs, in meeting this obligation, and the Commission will publish practical examples of how to comply.

Who has to comply

Almost everyone. The European Commission's AI Literacy Q&A says it concerns anyone in the organisation directly dealing with an AI system, and that people working for a contractor or service provider need the appropriate AI skills too. It also gives a plain example: staff using ChatGPT for ad copy or translation should be informed about specific risks, such as hallucination. Three concrete examples:

  • A 12-person law firm whose paralegals use Microsoft Copilot for first-draft contract review.

  • A 60-person e-commerce business whose customer service team uses an AI chatbot.

  • A 200-person SaaS company whose engineers use GitHub Copilot.

All three are "deployers" and all three need Article 4 measures.

What good measures look like in practice

The regulation does not define a curriculum, and the Commission says no mandatory training or test is imposed. A supervisor looking at your measures will reasonably ask three things:

  1. Can you show what you did? The law does not require records, but undocumented training leaves you with nothing to show. The Commission suggests keeping an internal record of trainings and other initiatives.

  2. Is it role-relevant? A generic "Intro to AI" video sent to 200 employees is weak if HR running CV screening, or finance running fraud detection, only got the same generic content. The Commission notes that different levels of training can be appropriate.

  3. Does it fit the AI systems actually in use? If your team uses ChatGPT and Copilot daily but the training only covers theoretical AI ethics, that is a gap. The Commission also warns that simply asking staff to read a tool's instructions is often ineffective.

The Commission's Q&A recommends at least four steps: a general understanding of AI, your organisation's role as provider or deployer, the risk of the AI systems you use, and measures tailored to your staff and the context of use. The CIPL Best Practices on AI Literacy suggest a similar layered approach: foundational AI literacy for everyone, role-specific modules for high-touch roles, and ongoing updates as your AI tooling changes.

The five things worth having on file

If a supervisor asked tomorrow what you have done under Article 4, these five documents answer the question. None of them require certification. They are evidence of your measures.

  1. AI usage inventory. A list of every AI-powered tool used in the company. ChatGPT counts. Microsoft 365 Copilot counts. The AI features in your HubSpot, Salesforce, or Notion count. If you do not know what AI you are deploying, you cannot train staff on it.

  2. Risk classification. For each tool, a short note: is it general-purpose (low risk), part of a high-risk use case under Annex III (HR screening, credit scoring, biometric ID), or in the prohibited-practices list under Article 5? Most SMEs find their inventory is overwhelmingly low-risk, and knowing that on paper is the point.

  3. Training programme. Foundational module for everyone, role-specific modules for staff who touch higher-risk systems. Aligned to the tools you actually use.

  4. Training records. A record per employee: name, modules completed, date completed. A spreadsheet is fine. A learning management system with certificates of completion is easier to keep up.

  5. AI usage policy. A short company policy document covering acceptable use, what data may and may not be put into AI tools, and an incident reporting procedure if an employee notices an AI giving harmful or biased output.

That is the core package. There is no certification, no accreditation, no audit body. The Commission says plainly that no certificate is needed. Article 4 is about measures and evidence, not credentials.

What enforcement actually looks like

This is the part most coverage of the AI Act gets wrong. The AI Act sets no specific fine for Article 4. Article 99 lists fines of up to €35 million or 7% of global turnover for prohibited AI practices (Article 5), up to €15 million or 3% for a list of operator obligations such as those of high-risk deployers under Article 26 and transparency under Article 50, and up to €7.5 million or 1% for misleading information to authorities. Article 4 is in none of those tiers. For SMEs and start-ups, the lower of the two figures in each tier is the maximum.

Instead, Article 99(1) requires Member States to set penalties and other enforcement measures for any infringement, which may include fines, warnings and non-monetary measures, and which must take the interests and economic viability of SMEs into account. National market surveillance authorities supervise Article 4 from 2 August 2026. The Commission says any sanction must be proportionate and might be more likely where there is proof of an incident caused by a lack of appropriate training and guidance.

In the Netherlands, the implementing law is still a draft and the Autoriteit Persoonsgegevens has no AI Act powers yet. Under the draft, the AP, AFM and DNB would supervise Article 4, and a shortfall would lead to an order to fix it within a set period (a "last onder dwangsom"), not a fine. We are not aware of any fine for Article 4 anywhere so far.

For most SMEs that translates into a real but bounded risk profile: an order to improve if your measures fall short, liability under national law if an untrained employee causes harm with an AI system, reputational damage, and contract questions as enterprise customers ask suppliers about their AI literacy measures in procurement.

Practical first step

If you have done nothing, the highest-leverage 30 minutes you will spend this quarter is taking a proper inventory of every AI tool actually in use across your company. You cannot train people on systems you do not know you have. Once that exists, the rest of the package is straightforward. Document your AI literacy measures this month.

If you would rather not build it from scratch, that is what we do: start with the free 2-minute compliance check to see where you currently stand.

References

  1. EU AI Act Article 4 (full text)

  2. Regulation (EU) 2026/1744 (Digital Omnibus on AI)

  3. European Commission AI Literacy Q&A

  4. CIPL AI Literacy Best Practices (May 2025)

  5. EU AI Act Annex III (high-risk uses)

  6. EU AI Act Article 99 (penalties)

More in this guide

Not sure where your company stands?

Answer 15 questions about how your team uses AI. You get a compliance score and the gaps to close first, as a PDF.

Am I compliant? Check in 2 minHow Khairos Comply works