AI literacy training under Article 4 of the EU AI Act, as short videos

Watch the video full size, with its transcript

This guide is for HR, L&D and compliance teams in organisations that use AI tools: a chat assistant, a translation tool, an AI feature inside business software. It summarises the obligation, then outlines a two-minute video that puts it into practice.

What Article 4 says

In one sentence: providers and deployers of AI systems must take measures to support the AI literacy of their staff and of the people who use those systems on their behalf.

• Who is covered. A “deployer” is any organisation using an AI system in a professional capacity. A company whose employees use an AI assistant to draft or summarise is one.
• Since when. Article 4 has applied since 2 February 2025 (Article 113). Supervision and enforcement rules apply from 3 August 2026, according to the European Commission.
• What changed in 2026. The Digital Omnibus on AI, in force since mid-July 2026, amended Article 4: AI literacy remains an obligation, but no specific or “sufficient” level is mandated any more.
• What is not required. No certificate and no dedicated AI officer. The Commission says an internal record of trainings is enough to keep track.
• What to adapt. Measures take into account people’s technical knowledge, experience, education and training, and the context the tools are used in.

What a good “AI at work” video covers

A video does not make anyone an expert. It gets your usage rules across, the ones that prevent incidents. Six scenes, 1 min 30 to 2 min:

1. A real situation. “Monday, 9 a.m. Sophie has to summarise a client report before noon. She opens the AI assistant.” Viewers recognise themselves within ten seconds.
2. What the tool does, and what it does not. It predicts plausible text; it checks nothing. An analogy: a brilliant intern who has read none of your files.
3. The data rule. What never goes in: client data, personal data, documents marked confidential. The approved tools, and which account to use.
4. The checking rule. Every figure, quote and legal reference is checked against the source before it is sent.
5. People decide. The tool suggests, a person decides and signs. Example: rejecting an application is never automated.
6. The recap and a contact. Three rules on screen, and the name of the person to ask.

Two formats work together: 16:9 for the intranet or the LMS, and a 9:16 45-second cut with one message, for Teams or the internal newsletter.

How to make it with EducPilot, in three steps

1. Upload your AI usage policy, or paste your list of approved tools and internal rules as text. The assistant can also rely on an official web page you point it to.
2. Approve the plan. Check that every rule in the video comes from your policy: each sentence links to the page of your document that supports it. Nothing is produced before you approve.
3. Correct by annotating, then share. A note pinned on the frame (“name the approved tool exactly”) rewrites that scene only. Export an MP4 in 1080p, 16:9 or 9:16, with burned-in subtitles, in English, French or Chinese.

What the video does not replace

• Your written policy, which the video points to.
• Deeper training for exposed roles: teams that configure or deploy an AI system need more than two minutes.
• Legal analysis: “high-risk” AI systems come with other obligations, for a lawyer to assess.

Sources

• Regulation (EU) 2024/1689 (AI Act), official text on EUR-Lex, Articles 4 and 113.
• European Commission, AI Literacy: Questions & Answers, checked on 5 October 2026 (application since 2 February 2025, enforcement from 3 August 2026, Digital Omnibus amendment, no certificate needed).
• European Commission, Repository of AI literacy practices.