What AI literacy means under EU AI Act Article 4

Article 4 matters because AI use is no longer limited to technical teams. Employees now use AI inside writing, research, analysis, support, sales, HR and operations work.

Companies should be able to show that people using AI on their behalf understand the tools in context: what the system can do, what it cannot do, what data should not be used, and where human review is required.

The Four Pillars of Literacy

A useful literacy program covers more than prompt examples. The team should be able to explain and apply:

• /01Capabilities and limitations: What can the system actually do, and where does it hallucinate?
• /02Operating principles: How does the specific AI tool work in the context of their daily tasks?
• /03Risk awareness: Understanding data privacy, bias, and security risks associated with AI usage.
• /04Responsible use: Knowing when AI should not be used and where human oversight is required.

Proportionate Evidence

Article 4 does not prescribe a universal certificate. The practical standard is proportional literacy: a marketing manager using ChatGPT for social media needs different guidance than a data analyst using Gemini for financial forecasting.

At The Unlearning School, we help organizations document literacy through role-based adoption programs, participant outputs, safe-use guidance and evidence that people can apply AI responsibly in real work.