articleSeptember 2026· 9 min reading time

What is AI literacy? The Council of Europe's answer

AI literacy. The law requires it when an organisation uses AI. But what does it actually involve? And what does it mean for you? This article is for everyone who uses AI at work: employee, manager or business owner.

The rule is in Article 4 of the EU AI Act. Paragraph 1 reads, word for word:

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.

By this, the law means five things:

1.Who has to act? The organisation you work for, if AI is used there for work. The law calls such an organisation a deployer. Think of the company, local authority, school or hospital where you work. Large or small makes no difference. If you run your own business where AI is used, you are the deployer. The obligation also applies to providers: the companies that develop AI systems and place them on the market. Most organisations are not providers. You can read the difference between a deployer and a provider in another article.

2.For whom? For you and your colleagues who work with AI. People who work with AI on the organisation's behalf, such as freelancers, are covered too.

3.What has to happen? The organisation takes measures that help people build knowledge and skills around AI. Think of explanations, training, practising with real situations or clear agreements about what is and is not allowed.

4.Tailored. Not everyone needs to know the same things. The organisation takes into account what someone already knows and can do, what AI is used for and the people affected by the outcome. Someone who has AI rewrite an email needs different knowledge from an HR officer who uses AI to screen job applicants.

5.No guarantee. The organisation does not have to guarantee that everyone reaches a certain level. It does have to genuinely do something. The law does not prescribe how it goes about that.

What is AI literacy?

Article 3 of the same law defines the term:

'AI literacy' means skills, knowledge and understanding that allow providers, deployers and affected persons, taking into account their respective rights and obligations in the context of this Regulation, to make an informed deployment of AI systems, as well as to gain awareness about the opportunities and risks of AI and possible harm it can cause.

In plain words: AI literacy means knowing and being able to do what it takes to use AI sensibly. You understand what AI can and cannot do. You see the opportunities. And you recognise the risks and the harm AI can cause.

In practice, this includes for example:

•knowing what AI is and roughly how it works, for example that a chatbot puts together a likely answer and does not check that answer itself;

•knowing which AI is used at work and what for;

•knowing what AI can get wrong, so you check the result yourself (see below);

•knowing which information you should not enter into an AI tool, such as personal data or confidential information;

•recognising when AI plays a part in decisions about people and what that means for them;

•knowing the agreements your organisation has made about AI.

Everything AI can get wrong

The best-known word for this is hallucination. But much more goes wrong than that. And it does not happen now and then. It happens regularly. AI can:

•make up facts, figures, names, sources or quotes that do not exist;

•bring things into the answer that you never mentioned;

•make its own assumptions and simply carry on with them;

•do something other than what you asked, or only part of it, and still say it is finished;

•ignore your instructions or forget them halfway through a long conversation;

•give outdated information without saying so;

•calculate or count incorrectly;

•agree with you, even when you are wrong;

•give a different answer to the same question each time;

•make the same mistake again, even after you have pointed it out.

And all in the same confident tone. A wrong answer sounds just as convincing as a correct one. So check what AI gives you before you use it. How thoroughly depends on what you are using it for and what a mistake could cost.

The European Commission itself gives an example. Anyone who uses a chatbot at work to write or translate texts should know that AI can make things up.

The Council of Europe and its recommendation

The EU AI Act does not say how deep that knowledge should go or exactly which topics it covers. The Council of Europe does describe it, in a framework of its own. That framework is not an explanation of the law. More on that below. On 2 September 2026, the Council of Europe adopted its recommendation on AI literacy. In the rest of this article, you can read what it contains and what you can do with it.

The Council of Europe is an organisation of European countries that stands up for human rights, democracy and the rule of law. It is separate from the European Union. All EU countries are members, as are other European countries such as Norway, Switzerland and the United Kingdom.

A recommendation from the Council of Europe is not a law. It is addressed to the governments of the member states and does not oblige organisations to do anything. The EU AI Act is legislation. Organisations do have to comply with that.

Even so, this recommendation carries weight. It was adopted by the Committee of Ministers, in which all member states are represented. It also fleshes out an article of the Council's AI treaty. That article asks the countries that are party to the treaty to promote digital literacy and skills.

The three sides of AI literacy

According to the Council of Europe's recommendation, AI literacy has three sides. Compare it with driving a car: you need to know how a car works, you need to be able to drive and you need to take other people on the road into account.

•Knowing how AI works. AI works with probabilities. A chatbot, for example, predicts which words fit an answer. It does not think like a human. It can be wrong, even when it sounds very sure of itself.

•Using AI well. You use AI safely and sensibly in your work. You know what it does well and above all where it goes wrong. Think of AI that makes things up, fills in something you never said or does something other than what you asked.

•Knowing what AI does to people. What does it mean for their rights, for their freedom to make their own choices and for democracy? Think of a customer who is turned down and wants to know why.

The Council of Europe puts extra emphasis on that third side. According to the Council, it is not only about knowing how the tools work. It is also about understanding what AI does to society. Other frameworks, according to the Council, focus mainly on technical knowledge.

Four principles from the recommendation

The Council of Europe's recommendation contains 22 principles. They build on the three sides you read about above. Four of them are useful for everyone who works with AI:

•AI is more than chatbots. It is often built into software you use every day, such as the spam filter in your mailbox or the suggestions in an online shop.

•AI is a tool, not a human being. It can sound very human and politely say "Great question!" Still, do not trust it blindly.

•Promises about AI need backing up. Think of a supplier who promises that their tool is safe or saves a lot of time. They must be able to prove it with independent evidence.

•Keep using your own judgement. "The computer said so" is no excuse. Do not just accept what AI gives you. You do not hand your judgement and your responsibility over to AI.

Note: the recommendation does not explain the law

The Council of Europe's recommendation refers to Article 4 of the EU AI Act. But it does not explain what that article means. Nor does it say what an organisation must do to comply with Article 4. What the law requires is in the law itself.

You can, however, use the Council of Europe's recommendation as a yardstick. Hold your own knowledge up against it. Or your organisation's approach. That way you can see where something is still missing. You can read exactly what Article 4 requires of an organisation in what Article 4 means in practice for an organisation.

Three questions to ask yourself

Ask yourself these three questions:

1.Do you know that AI makes mistakes? And what about your colleagues?

2.Do you know which AI tools are used at work and what for?

3.Is anyone talking about what AI does to people? For example, does someone affected by a decision involving AI know what happened and who they can turn to?

If you answer no to any of them, you know where to start. On your own or together with your organisation.

Are you helping to organise AI literacy in your organisation? These articles on our site can help:

•Organising a training course alone is not enough: why AI literacy is more than a tick on a checklist

•A step-by-step plan: our guide in five phases

•Short lessons: why microtraining works

•How an organisation manages AI: our overview of AI governance

Do you work in government or education?

In its recommendation, the Council of Europe asks member states to make sure that public officials and other professional groups can take training in AI literacy. Education staff receive particular attention.

If you work for a local authority, read what the EU AI Act asks of local authorities. If you work in education, read what the EU AI Act means for schools.

Getting started

This part is about what we offer. It is separate from the Council of Europe's recommendation.

Our AI literacy training is available by role and by sector. There is also separate training for public authorities and training for education.

Would you like to know which AI chatbots are used in your organisation? The free AI Adoption Scan maps this with an anonymous questionnaire. Do you already use AI but get too little out of it? That is what Effective AI at Work is for. Would you like a broader overview of your organisation, including a risk assessment? Then there is the AI Adoption Assessment.

Questions? Get in touch with us.

Sources

Recommendation CM/Rec(2026)12 of the Committee of Ministers to member States on artificial intelligence literacy, Council of Europe

New Council of Europe recommendation places education at the heart of response to AI, Council of Europe

Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law, Council of Europe

Regulation (EU) 2026/1744 (Digital Omnibus), with the current text of Article 4, EUR-Lex

Regulation (EU) 2024/1689 (AI Act), with the definition in Article 3, EUR-Lex

AI Literacy - Questions & Answers, European Commission

Written by Rob Ummels in collaboration with Claude (Anthropic). Editorial responsibility: AIAdopt.

What is AI literacy? The Council of Europe's answer

What is AI literacy?

Everything AI can get wrong

The Council of Europe and its recommendation

The three sides of AI literacy

Four principles from the recommendation

Note: the recommendation does not explain the law

Three questions to ask yourself

Do you work in government or education?

Getting started

Sources

You'll want to know this too...

Do I need to label AI? 7 everyday situations from Article 50

AI in healthcare: what the EU AI Act changes in practice for healthcare organisations

AI in education: what the EU AI Act means in practice for schools and educational institutions

AI literacy for employees: why microtraining works

What does AI training actually deliver? The cost of doing nothing

Does an ISO 42001 Awareness Training satisfy the AI literacy duty under the EU AI Act?

The Digital Omnibus changes the EU AI Act. But not in the way you think.

Your organisation has an AI policy. But do you know if anyone follows it?

The EU AI Act for local authorities: what must you already have in place?