Best AI Training for Employees

AI literacy trainingwhat the EU AI Act’s Article 4 asks

Article 4 of the EU AI Act asks companies that use AI to support the AI literacy of their staff. This guide sets out what the text says after its 2026 amendment, what the European Commission says it means and how a mid-sized company can meet it.

Contents
  1. The question behind the search
  2. What Article 4 says
  3. What changed in 2026
  4. What the European Commission says it means
  5. When and how it is enforced
  6. Which training meets it
  7. A plan for a company of 50 to 300 people
  8. Sources

A company whose staff use ChatGPT, Claude or Copilot, and which sells to or employs people in the European Union, has probably heard that AI training is now required. The rule is Article 4 of the EU AI Act. It is short, it was amended in 2026, and it is often described as stricter than it is. This guide reads it from the text and from the European Commission's own answers. It is not legal advice.

What Article 4 says

Article 4 applies to providers and deployers of AI systems. A company that uses an AI system in its work is a deployer. As amended, its first paragraph says that providers and deployers "shall take measures to support the development of AI literacy" of their staff and of other persons dealing with the operation and use of AI systems on their behalf, taking into account their technical knowledge, experience, education and training, the context in which the systems are used and the people they are used on. It adds that this obligation does not require them to guarantee any specific level of AI literacy of any individual⁠1.

The Act defines AI literacy as the skills, knowledge and understanding that allow providers, deployers and affected persons to make an informed deployment of AI systems and to gain awareness of the opportunities and risks of AI and the harm it can cause⁠1.

What changed in 2026

The original text asked companies to take measures to ensure, "to their best extent, a sufficient level of AI literacy" of their staff⁠1. The Digital Omnibus on AI changed that. According to the European Commission, the amendments entered into force in mid-July 2026: AI literacy remains an obligation for providers and deployers, but no specific or "sufficient" level is mandated; the Commission and the Member States continue to support companies, including by publishing practical examples of how to comply; and the AI Board must adopt recommendations on AI literacy⁠2. The amended second paragraph names small and medium-sized enterprises as a group the Commission and Member States should support⁠1.

What the European Commission says it means

The Commission's questions and answers, last updated on 20 October 2026, settle most of the practical points⁠2.

QuestionThe Commission's answer
Does a company whose staff use ChatGPT need to comply?Yes. For employees using it to write advertising text or translate, they should be informed about specific risks, such as hallucination.
Is a certificate needed?No. Organisations can keep an internal record of trainings and other guiding initiatives.
Is an AI officer or governance board required?No specific governance structure is mandated.
Is there a mandatory format?No. There is no one size fits all, and no strict requirements or mandatory trainings are imposed.
Who counts as staff?Other persons can include contractors, service providers and clients under the organisation's remit.
Does it reach companies outside the EU?It applies to actors inside and outside the EU when an AI system is placed on the EU market, used in the EU or affects people located in the EU.
Is relying on the tool's instructions enough?Generally not; further measures are necessary.

The Commission suggests four elements as a starting point: a general understanding of AI within the organisation, the organisation's role as provider or deployer, the risks of the AI systems it uses, and actions tailored to the staff's knowledge and the context of use, covering the legal and ethical aspects as well⁠2. Deployers of high-risk systems carry an additional duty under Article 26 to ensure staff are trained for human oversight⁠2.

When and how it is enforced

Article 4 has applied since 2 February 2025. Supervision and enforcement start on 2 August 2026, and they sit with national market surveillance authorities, not with the EU AI Office. Penalties are set by national law, and the Commission says enforcement should be proportionate, taking into account factors such as the gravity of an infringement and whether it was intentional or negligent⁠2.

A compliance lead and a team leader mark a printed list at a high table in an office.
A dated list of who completed which training, kept internally, is what the Commission describes.

Which training meets it

Because the law sets no format, the choice is about fit, not compliance. What matters is that the training covers the AI systems the company actually uses, their risks and the company's rules, and that the company can show what it did.

  • A baseline for everyone. Anthropic's free AI Fluency course runs about four hours and covers delegation, description, discernment and diligence⁠3; its course on AI capabilities and limitations covers what large language models can and cannot do⁠3.
  • Tool-specific training. Companies on Microsoft Copilot can use Microsoft's free material and live Digital Accelerator cohorts⁠45; OpenAI Academy runs courses for knowledge workers on prompting, review and repeatable workflows⁠6.
  • Records and certificates. Section's team plans include certificate programs and an admin dashboard⁠7, and Coursera for Teams includes skill tracking and certificates⁠8. The Commission does not require certificates, but a dashboard makes the internal record easier to keep.
  • Training tied to the company's own systems. OpenLazarus offers AI training for employees tied to building and governing the company's own AI teammates, with permissions for data, files, messages and spending set as Always, Ask or Never⁠9. That suits the Commission's point that the risks of the systems actually in use belong in the training.

A plan for a company of 50 to 300 people

  1. List the AI systems in use. Approved tools, and the ones staff use anyway. BCG found that more than half of employees without the tools they need find alternatives⁠10.
  2. Write short rules. Which data may go into which tool, and what needs a person's approval.
  3. Run a baseline course for everyone, contractors included. Choose one that covers how AI works and where it fails, such as hallucination.
  4. Add role-specific sessions. Teams that use AI on customers, money or personal data need training on those systems and their risks.
  5. Keep the record. Who completed what, when, and which systems it covered. Review it when tools change.

The end state is a one-page AI literacy record: the systems in use, the rules, the training each group received and the date. It answers a regulator's first question, and it is the same document a leadership team needs to manage AI anyway.

For the providers that deliver this training, see the comparison of AI training for employees.

Sources

  1. Article 4: AI literacy. artificialintelligenceact.eu. Accessed 19 Nov 2026.
  2. AI Literacy: Questions & Answers. European Commission. Accessed 19 Nov 2026.
  3. Claude Academy. Anthropic. Accessed 19 Nov 2026.
  4. Microsoft Copilot. Microsoft Adoption. Accessed 19 Nov 2026.
  5. Digital Accelerator. Microsoft Adoption. Accessed 19 Nov 2026.
  6. OpenAI Academy. OpenAI. Accessed 19 Nov 2026.
  7. Pricing. Section. Accessed 19 Nov 2026.
  8. Online Training for Teams. Coursera. Accessed 19 Nov 2026.
  9. OpenLazarus. OpenLazarus. Accessed 19 Nov 2026.
  10. AI at Work 2025: Momentum Builds, but Gaps Remain. Boston Consulting Group, 26 Jun 2025. Accessed 19 Nov 2026.

Updated 19 Nov 2026: First published, with Article 4 as amended by the Digital Omnibus on AI.