AI Literacy for Staff: EU AI Act Article 4 (Team Pack)
Understand what AI tools can and cannot do, use them safely with company data, check their output and spot uses the EU AI Act restricts, so your organisation can show its Article 4 measures.
What you will learn
- Explain what Article 4 requires, since when, from whom, and how the 2026 Digital Omnibus changed it.
- Describe in plain words how AI and generative AI work and why they make errors such as hallucinations and bias.
- Recognise the AI Act risk levels and identify prohibited and high-risk uses that concern your job.
- Use generative AI tools at work without exposing confidential or personal data.
- Check AI output, keep a human in control and stay accountable for decisions about people.
- Contribute to your team’s AI literacy measures: tool inventory, rules, records and reporting of problems.
Modules
- Article 4: what the law asks of your organisation
- How AI works and why it gets things wrong
- Risk levels: what is banned, what is high-risk
- Using generative AI safely with company data
- Keep a human in control and stay accountable
- Your team’s Article 4 measures in practice
Course facts
- Level
- Awareness
- Estimated study time
- about 45 minutes
- Final assessment
- 24 questions · Pass mark 80%
- Available in
- English, Français
- Price
- Free. Optional verifiable certificate: $5
- Last updated
- Published by
- AMAADOR ACADEMY
Who this course is for
All employees, managers, contractors and interns who use AI tools at work (chat assistants, copilots, translation, HR or customer-service software), in organisations that provide or deploy AI in the EU or whose AI output affects people in the EU, including firms in Morocco and the GCC serving EU clients.
Prerequisites
None. You only need to use a computer or smartphone at work.
Course content
1. Article 4: what the law asks of your organisation
The EU AI Act is Regulation (EU) 2024/1689. It entered into force on 1 August 2024 and applies in stages. Two parts came first, on 2 February 2025: the bans on certain AI practices (Article 5) and the AI literacy duty (Article 4).
The original Article 4 asked providers and deployers to take measures to ensure, to their best extent, a sufficient level of AI literacy of their staff. In July 2026 the EU adopted Regulation (EU) 2026/1744, the Digital Omnibus on AI, which replaced Article 4. The duty did not disappear. It was reworded.
| Point | Article 4 today (as amended) |
|---|---|
| Who | Providers and deployers of AI systems |
| Duty | Take measures to support the development of AI literacy of staff and others operating or using AI on their behalf |
| Tailored to | People’s technical knowledge, experience, education and training, the context of use, and the people the AI is used on |
| Not required | Guaranteeing a specific level of AI literacy of any individual |
| Support | Commission and Member States help, especially SMEs; the Commission publishes practical examples; the AI Board adopts recommendations |
2. How AI works and why it gets things wrong
The AI Act defines an AI system (Article 3(1)) as a machine-based system that operates with some autonomy and infers from its input how to generate outputs such as predictions, content, recommendations or decisions. The key word is infer: the system was not given every rule; it learned patterns from data.
| Type | What it does | Workplace examples |
|---|---|---|
| Classic rules software | Follows rules written by people | Payroll formulas, a spreadsheet macro |
| Machine learning | Learns patterns from past data to predict or classify | Spam filter, fraud alerts, CV ranking, demand forecasting |
| Generative AI | Produces new text, images, audio or code | Chat assistants, copilots, meeting summaries, translation |
A large language model (LLM), the engine of chat assistants, is trained on huge amounts of text to predict the next likely word. It does this so well that its answers sound expert. But it does not look facts up unless it is connected to a search or document tool, and it has no built-in sense of truth.
Three ideas help you use these tools well:
3. Risk levels: what is banned, what is high-risk
The AI Act regulates uses, not technologies. The same model can be harmless in one use and banned in another. Knowing the levels helps you spot when a request needs to go to your manager or compliance team.
| Level | Meaning | Examples |
|---|---|---|
| Unacceptable (Article 5) | Prohibited since 2 February 2025 | Social scoring; manipulative or deceptive techniques causing significant harm; exploiting vulnerabilities; emotion recognition in the workplace or in education (except for medical or safety reasons); biometric categorisation to infer sensitive traits; untargeted scraping of facial images |
| High-risk (Article 6, Annexes I and III) | Allowed with strict requirements | AI for recruitment, selection, promotion, dismissal, task allocation and performance monitoring; credit scoring; access to education; safety components of products |
| Transparency (Article 50) | People must be told | Chatbots must say they are AI; deepfakes must be disclosed; AI-generated content marked |
| Minimal | No specific AI Act rule beyond Article 4 | Spam filters, spell checkers, most drafting help |
Dates matter. The bans apply now. The Digital Omnibus (Regulation (EU) 2026/1744) moved the start of the high-risk rules: 2 December 2027 for Annex III uses such as employment, and 2 August 2028 for AI in products covered by Annex I. Article 4 does not wait for those dates.
4. Using generative AI safely with company data
Everything you type or upload into an AI tool leaves your hands. Depending on the tool and contract, it may be stored, reviewed by the provider, used to improve the model or processed outside the EU. The question is not “is the tool clever?” but “is this tool approved by my employer for this kind of data?”.
| Data class | Public or personal free tool | Company-approved AI tool |
|---|---|---|
| Public information (published website text, public laws) | Usually fine | Fine |
| Internal (procedures, non-sensitive notes) | No | Yes, per policy |
| Confidential (contracts, prices, source code, strategy, client files) | No | Only if the policy allows it for that tool |
| Personal data (names, emails, CVs, health, payroll) | No | Only with a legal basis, minimised, per policy |
| Secrets (passwords, API keys, access tokens) | Never | Never |
Personal data brings in the GDPR: purpose limitation, data minimisation and security apply when you paste a client’s complaint or a colleague’s appraisal into a prompt. In France the CNIL publishes guidance on AI; in Morocco, Law 09-08 and the CNDP play that role for personal data.
5. Keep a human in control and stay accountable
The biggest everyday risk is not a rogue robot. It is automation bias: the habit of trusting a machine’s answer because it is fast, tidy and usually right. The AI Act names this risk in Article 14 on human oversight: the people overseeing high-risk AI must stay aware of the tendency to over-rely on its output.
Match your checking effort to the stakes:
| Stakes | Examples | Minimum check |
|---|---|---|
| Low | Rewording an internal note, brainstorming titles | Read it once; does it say what you mean? |
| Medium | Client email, report summary, translation sent outside | Check facts, figures, names and tone against the source |
| High | Legal, financial, safety, medical or HR content; anything about a named person | Verify every claim in the original source; second human review; record that AI was used |
- Compare — put the output next to the source document or data.
- Check the numbers — recalculate totals, dates and percentages yourself.
- Open the references — every law, study or link must exist and say what is claimed.
- Read for people — look for unfair or stereotyped statements about any group.
- Decide — use, edit or discard. If you cannot verify it, do not use it.
6. Your team’s Article 4 measures in practice
Article 4 asks for measures suited to people’s knowledge and to the context of use. A team can build them in a few steps, drawing on the approach in the Commission’s Q&A.
- List the AI in use — tools bought, AI features switched on in existing software, and what people actually use.
- Clarify the role — for each tool, is the organisation a provider or a deployer?
- Rate the use — minimal, transparency, high-risk, or a practice to stop (module 3).
- Define levels by role — what each group must know.
- Deliver and record — this course, team briefings, guides, practice sessions.
- Review — when a tool, a use or the law changes.
| Group | Typical AI use | Suggested level |
|---|---|---|
| All staff | Chat assistant, translation, summaries | This awareness course plus the team AI rules |
| Managers | Approving AI use, reviewing AI-assisted work | Plus risk levels, HR and decision-making limits |
| HR, recruiters | Screening, performance tools | Plus high-risk duties, human oversight, GDPR rights |
| IT, developers | Coding assistants, building AI features | Plus security, data flows, provider duties if building |
| Customer-facing staff | Chatbots, AI-drafted replies | Plus transparency and escalation to a human |
The full lesson continues in the interactive course on this page.
Sources and references
- Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (AI Act)
- Regulation (EU) 2026/1744 (Digital Omnibus on AI), amending the AI Act, including Article 4
- AI literacy: questions and answers
- Living repository to foster learning and exchange on AI literacy
- Artificial intelligence: CNIL guidance and resources (in French)
- Regulation (EU) 2016/679, General Data Protection Regulation (GDPR)
- AI Risk Management Framework (AI RMF 1.0)
Education only, not legal advice. Based on EU law: Regulation (EU) 2024/1689 (AI Act) as amended by Regulation (EU) 2026/1744, with pointers to the GDPR and French labour law. Your employer’s AI policy and the official texts on EUR-Lex prevail. Completing this course is one of the measures an organisation can record under Article 4; it is not a certification and does not on its own prove compliance.
All courses · Cloud & DevOps - Free courses, certificate $5 | AMAADOR ACADEMY