AI literacy training: what employees need to know (and a 30-minute session)

The EU AI Act asks organisations to support AI literacy. It does not ask them to certify it. That one point is easy to miss, and it shapes how you plan training. You have to take measures that fit your people and the way they use AI, but no certificate and no set level is required.

AI is now everyday work, which is why this matters. In 2025, 19.95% of EU enterprises with 10 or more employees used at least one AI technology, up 6.47 percentage points on 2024 (Eurostat: Use of artificial intelligence in enterprises). About one in five, and rising.

This post gives you the 5 core skills to teach, a guide to how deep to go by role, a timed 30-minute session and a simple way to show it happened. It covers EU rules only and is general information, not legal advice.

What the EU AI Act asks of employers

Article 4 applies to providers and deployers of AI systems. They must take measures to support the AI literacy of their staff and of other people who operate or use AI systems on their behalf. The amended text says the measures should take into account people's "technical knowledge, experience, education and training" and the context the AI systems are used in (AI Act Service Desk: Article 4). The Act defines AI literacy in Perkara 3 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 possible harm it can cause.

Here is what it does not ask for. Article 4 was amended by the Digital Omnibus on AI, and the text now says the obligation "does not require providers or deployers to guarantee any specific level of AI literacy of any individual". The Commission's own Q&A adds that there is no obligation to measure staff knowledge, and that no certificate is needed: organisations can keep an internal record of training and other initiatives (European Commission: AI literacy Q&A).

It also says that simply asking staff to read a tool's instructions for use is often ineffective. Enforcement sits with national market surveillance authorities, and supervision and enforcement rules apply from 2 August 2026. So the practical aim is a documented, sensible effort that fits your people, not a pass mark.

The 5 core skills

These stay the same whichever tools your team uses. Teach them tool-neutral, then add examples from your own workplace.

1. Know what the tools are and what they cannot do. AI tools can be wrong and sound sure. Staff should know that an answer which reads well is not the same as an answer that is right. The Commission's Q&A names hallucination as a risk people using a generative tool should be told about.
2. Give clear instructions and check the output before you use it. Say what you want, who it is for and what to leave out. Then check names, numbers and claims against a trusted source before the result goes to a customer, a manager or a document.
3. Protect people's data and company information. Know what must never go into a tool: personal data, customer details, contracts, anything marked confidential. If you are unsure, treat it as off limits until you have asked.
4. Know your organisation's rules and who to ask. Which tools are approved, what they may be used for and who answers questions. Our AI acceptable use at work template is a ready-made start for this skill.
5. Spot risks and report problems. A biased result, a made-up source, a tool doing something it should not. People need to know a report is welcome and where to send it.

Our pick: skill 2, check the output. It is a habit that prevents many common mistakes, and it works the same for any tool. If you can only teach one thing this month, teach people to verify before they send.

For a wider view of how AI changes the learning side of work, see what AI changes about workforce learning.

How deep to go by role

The Commission says different levels of detail are allowed, and that measures should reflect what people already know and the context they work in. A simple split:

Where your organisation deploys high-risk AI systems, the Commission says additional measures may be relevant. Staff who oversee those systems must be sufficiently trained for human oversight under Article 26. Treat that as a separate, deeper track.

A 30-minute session plan

There is no set length for AI literacy training, so a 30-minute session is a practical first step, followed by refreshers. This plan uses five timed blocks and no vendor tools, so you can run it with any team.

Sediakan: a short poll, one made-up example of AI output with an error planted in it, and a one-page summary of your organisation's rules.

1. Minutes 0 to 5: opening poll. Ask "What do you already use AI for at work?" Use the answers to pitch the rest of the session.
2. Minutes 5 to 10: how these tools work and fail. Explain in plain words that they predict likely text and can be wrong with full confidence. Show one failure.
3. Minutes 10 to 20: try and check. Hand out the made-up example. Pairs find what is wrong with it and say what source they would check. Take two or three answers.
4. Minutes 20 to 25: your organisation's rules. Walk through the one-pager: approved tools, data that stays out and who to ask.
5. Minutes 25 to 30: questions and a commitment. Open the floor to anonymous questions, then ask everyone to write one thing they will do differently this week.

Our pick: the try-and-check block. Finding an error yourself tends to stick better than being told about it, and the block practises skills 1 and 2 in one go.

Measuring it

Because no certificate or level is required, measure what helps you improve. With about 20% of EU enterprises using AI in 2025 and the share growing, more of your people meet these tools every year, so track habits rather than scores.

1. Kehadiran. Who came, and who still needs a slot. Keep the list: the Commission says an internal record is enough.
2. A 5-question check at the end. Use it to find gaps, not to pass or fail anyone.
3. A confidence rating before and after. "How confident are you checking AI output?" from 1 to 5.
4. A short follow-up after 30 days. Ask what people have used and what went wrong.
5. Reports raised. More questions and reports after training is a good sign.

Here is a 5-question check you can use as it stands. Each answer comes first.

1. An AI tool gives you a confident answer with a statistic. What do you do first? Check the statistic against a trusted source before you use it.
2. Can you paste a customer's details into a tool to speed up a reply? Only if your organisation's rules say that tool is approved for that data. If not sure, ask first.
3. Does a well-written answer mean a correct one? No. AI tools can sound sure and still be wrong.
4. Who do you tell if a tool produces something biased or made up? The person or channel named in your organisation's rules.
5. Does the EU AI Act require a certificate for AI literacy? No. The Commission says no certificate is needed, and the obligation does not require a specific level of AI literacy for any individual.

Pilihan kami: soalan 1. It tests the habit from skill 2 in one line, and it is a question that is easy to rush.

Ways to host AI literacy training with AhaSlides

This session works best when people answer rather than listen. Use these slide types.

1. Pilih Jawapan for the 5-question check, or to ask "which of these prompts is unsafe?" and "what is wrong with this output?"
2. Awan Perkataan: "How do you use AI at work today?" You see the real tools and tasks before you teach.
3. Skala Penilaian: "How confident are you checking AI output?" Run it before and after, and compare.
4. Terbuka Berakhir: "What task would you like help with?" The answers shape your next session.
5. Soalan & Jawapan, anonymous, so people ask the questions they would not ask out loud.

The EU: AI literacy (EU AI act article 4) course template gives you the slides ready to adapt. For more on what to teach, see top employee training topics for 2026, and for the privacy side of AI use, AI documentation privacy training.

Soalan lazim

What is AI literacy?

The skills and understanding to use AI tools sensibly, knowing their benefits, risks and limits. The AI Act defines the term in Article 3: 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 possible harm it can cause.

Is AI literacy training mandatory?

Under the EU AI Act, providers and deployers must take measures to support AI literacy, but the Commission says no certificate or specific level is required. The measures should fit your people and the context. This applies to EU rules only; check the rules that apply to your own organisation.

Who needs AI literacy training?

Staff and others who operate or use AI systems on the organisation's behalf, with depth matched to the role. The Commission says this can include contractors and service providers, not only employees.

How long should AI literacy training take?

There is no set length. A 30-minute session is a practical start, followed by refreshers as tools and rules change. Staff in deeper roles, such as those overseeing high-risk systems, will need more.

What should an AI literacy session cover?

How the tools work and fail, how to check outputs, how to protect data, your organisation's rules and how to report a problem. The 5 core skills above cover each of these.

Di mana untuk bermula

Book one 30-minute session for the people who use AI most, run the try-and-check block and keep the attendance list. Then add a 30-day follow-up.