AI literacy training for employees: how to organise it

AI literacy training for employees is organised as follows: identify who in the company uses artificial intelligence, prepare concise practical training materials, conduct a knowledge check and record who completed the training. This is not a multi-day course - for a small company, a focused programme and a record of participation can be useful, with duration chosen to suit the roles and AI use. The duty to take measures supporting literacy arises from Article 4 of the EU AI Act (Regulation (EU) 2024/1689) and applies equally to Lithuanian micro-enterprises (1 - 30 employees) as it does to large corporations.

Good training starts not with slides, but with the actual tools the company uses: ChatGPT, Copilot, translation programmes, image generators or accounting automation.

How does employee training support the AI literacy obligation?

Article 4 of the EU AI Act obliges every organisation that uses AI systems to take measures to support the development of AI literacy among its employees and other persons using AI, taking into account their knowledge, experience, training, context and the persons who may be affected. This obligation has applied since 2 February 2025 - earlier than most other requirements of the regulation. Note: the purpose of the measures is to support literacy development, but there is no requirement to guarantee a specific or sufficient individual level.

Three key points to understand:

• Applies to small companies too. Article 4 does not exempt any organisation based on the number of employees or turnover. If your accountant uses AI to process accounts and your marketing specialists use it to generate texts, the obligation applies.
• Buying a tool is not enough. Literacy means that employees understand how AI works, what risks it poses and how to use it responsibly - not just knowing how to press a button.
• Keep useful evidence. Records can help explain what was taught and who took part; Article 4 does not prescribe a particular document or training format.

If you are still unsure what this concept exactly means, start with the article AI literacy - what it is and who needs it, then return to organisation.

Where to start: AI usage inventory

Before training employees, you need to know what they are actually using. In many small companies, AI tools appear "on their own" - one employee registers for ChatGPT, another enables Copilot, a third uses a translation tool. This is called shadow AI and often remains invisible to management.

Start with a simple list. Ask your team three questions:

1. Which AI tools do you use at work? (ChatGPT, Gemini, Copilot, Midjourney, translation, transcription tools, etc.)
2. For what tasks? (writing texts, coding, customer responses, data analysis)
3. Do these tools process client or personal data?

This inventory later becomes your AI tools register and also indicates which topics are most important in training. For example, if the team heavily uses generative chatbots, pay more attention to data confidentiality and hallucination recognition.

Practical tip: conduct the inventory not as an official audit, but as an open conversation. Employees often fear admitting they use a tool that management does not know about, so they stay silent. Clearly state that the goal is not to punish, but to establish rules, and you will get a more accurate picture. For a team of five, this conversation takes half an hour, but saves many mistakes later.

What topics should company AI training cover?

A good AI literacy programme for a small company covers four topic groups. The table below shows the minimum content worth including.

A more detailed programme plan with modules and durations can be found in the article AI training programme for companies. Practical daily rules for employees - something they can print and keep at hand - are summarised in the Employee AI Guide.

A small company does not need to fit everything into a single event. It is more convenient to divide the material into short, self-study modules that the employee can absorb over a few sessions. The secret to specificity is to provide at least one example from your own activities in each topic. If you work in e-commerce, show what an inadequately AI-generated product description looks like; if you provide services, explain why a client’s contract cannot be pasted into a public chatbot. Such examples are remembered better than general rules.

How to organise training step by step

A small company does not need a training platform or an external lecturer. The most practical approach is as follows:

1. Prepare short material (1 - 2 hours of reading). Cover the four topic groups from the table. Write specifically, with examples from your activities, not abstractly.
2. Assign a responsible person. Even if it is the manager themselves - someone must be accountable for ensuring that the training takes place and is recorded.
3. Schedule the time. For new employees - during the first week; for existing employees - a one-off session, followed by a refresher at least once a year.
4. Conduct a knowledge check. A short test (5 - 10 questions) shows whether the material is understood and also creates objective evidence.
5. Record completion. A certificate with the date and employee’s name can document the activity; it does not certify full legal compliance.
6. Record the data. A list of "who, when, what completed" is your compliance document.

After these steps, it is worth checking whether the training achieved its goal. How to do this objectively is explained in the article how to measure AI literacy - mere participation is not enough; the level of understanding is also important.

How often should training be repeated?

The AI field changes rapidly, so one-off training is not enough. A practical rule for small companies:

• Initial training - for every employee who uses or starts using AI.
• Onboarding for new employees - training during the first days of work, as part of onboarding.
• Annual update - a short session reviewing new risks, new tools and changes in company policies.
• Ad hoc - when the company starts using a new significant AI tool or when legal acts change.

Remember the regulation deadlines: general-purpose AI model rules began applying from 2 August 2025, subject to transitional provisions; Article 50 transparency rules applied from 2 August 2026, and most of the remaining obligations apply from 2 August 2026 (e.g., high-risk AI system requirements - from 2 December 2027, and high-risk AI embedded in regulated Annex I products - from 2 August 2028). Annual updates help you keep up with these stages.

What mistakes do small companies make?

The most common organisation mistakes to avoid:

• Training only "on paper". An email with a PDF is not training if no one reads it and no one checks understanding.
• No records. Training took place, but nothing was recorded, making it harder to explain its content and participants later.
• Too generic material. Universal slides without examples from your activities are of little use to employees.
• One-off event. Trained once and forgotten - without repetition, literacy fades.
• Ignoring shadow AI. Only "official" tools are covered in training, while ten others are actually used.

More similar pitfalls and how to avoid them are described in the list of all topics - see list of all articles.

How to prove that training took place?

From a compliance perspective, the most important thing is not the event itself, but the evidence. A minimal compliance package for a small company looks like this:

• AI usage policy (rules on how employees may use AI);
• AI tools register (which tools are used and for what);
• training materials;
• knowledge check results and certificates;
• compliance record - who, when, what completed.

Important: The EU AI Act does not require a specific certificate, a fixed number of hours, an external auditor or a dedicated AI officer. Documents and training are recommended practical evidence, but they do not guarantee legal compliance. Note that GDPR does not impose a blanket ban on AI tools; instead, lawful basis, data minimisation, use of approved tools and safeguards must be assessed case by case.

Manually preparing these documents takes time and requires a legal context. nebedirbam.lt automatically prepares the entire package for a fixed price of €149 + VAT (€299) - AI usage policy, tools register, compliance record, employee AI guide and AI literacy training with knowledge checks and certificates. Documents are generated automatically from templates reviewed by a compliance specialist, using your company’s answers. You can learn more about this on the homepage.

If you first want to try the training yourself without the full document package, there are also free AI literacy training courses - a good starting point before moving to full compliance. The free path is for one signed-in participant; team training is included in the paid documentation package.

Brief summary

AI literacy training for employees in a small company is a manageable, one-off project with annual updates, not a constant burden. Start with what the team actually uses, prepare specific material covering the four topic groups, conduct a knowledge check, issue certificates and record the data. The record helps explain the training that actually took place. Neither the record nor attendance alone establishes compliance with Article 4.

You can find the official text of the regulation on the EUR-Lex portal. For the latest information from the European Commission on the AI Act and literacy requirements, see the AI Act Service Desk and the Digital Strategy page. These sources provide the official interpretation of Article 4 of the EU AI Act and its application deadlines, including 2 February 2025 (initial application), 2 August 2026 (general rules, e.g., Article 50) and 2 December 2027 and 2 August 2028 (high-risk AI deadlines).