Most organisations in Norway already use artificial intelligence. Some have bought Microsoft Copilot, others use the free version of ChatGPT, and many have AI built into tools they have used for years – from accounting software to the customer service chat. What is often missing is a plan for how employees should use these tools. That is exactly what AI literacy is about.
What is AI literacy?
AI literacy is the knowledge employees need to use AI tools in a useful, safe and responsible way. It is not about programming. It is about understanding what the tool does, where it fails, and what should never be shared with it.
An employee with good AI literacy knows that a language model predicts the most likely answer. They therefore also know that the answer can be wrong, even when it sounds convincing. They know how to ask questions that produce better answers, and how to check the result before it is sent on to a customer.
What does Article 4 of the EU AI Act say?
The EU AI Act has a dedicated provision on AI literacy in Article 4. It says that organisations that develop or deploy AI systems must take measures to support the competence of staff and others who use those systems on their behalf. The requirement applies regardless of industry or size, and also covers ordinary, low-risk tools.
In Norway, the regulation will be implemented through a dedicated AI act. As of October 2026 the act has not been passed, and the government has indicated that the bill will go to parliament in 2027. But the content of the requirement is known, and Norwegian organisations with customers or activities in the EU already face it. What you do now will not be wasted.
Why it matters even without a legal requirement
Even without a legal requirement, there are three good reasons to prioritise AI literacy:
- Data leaks. Employees who paste customer lists, contracts or health data into a free AI tool may breach both privacy law and confidentiality duties.
- Errors that spread. AI that invents sources, figures or legal references can end up in reports, proposals and emails if nobody checks.
- Lost value. Many employees give up on AI after a few poor answers. With a little training they save real time every day.
What should employees know?
Good basic training covers at least the following:
- What AI is, and why it can be confidently wrong
- How to write clear requests that produce good, verifiable answers
- What should never be pasted into an AI tool – personal data, passwords, internal documents
- Which tools the organisation has approved, and why
- How AI can be misused for fraud, for example with cloned voices and hidden instructions in documents
Training should be adapted to the role. A case officer writing decisions needs different examples from a salesperson or a manager.
Three steps to get started
1. Map current use. Find out which AI tools are actually in use – including unapproved ones. A short, anonymous survey often gives surprising answers.
2. Write simple guidelines. One page is enough: which tools are approved, what may be shared, and who to ask when in doubt.
3. Give every employee practical training – and document it. Use employees' own tasks in the training. Keep an attendance list so you can show what has been done if customers, auditors or regulators ask.
Training nobody remembers isn't training. The best AI training uses employees' own tasks.
Summary
AI literacy is becoming a baseline requirement, in the same way as training in information security and privacy. Organisations that start now get both safer use and more value from the tools they already pay for. Start with a mapping exercise, write simple guidelines and give every employee short, practical training. A good first session for the whole workplace is our workshop AI at work.
This article is general information, not legal advice.