Reading time: 5 minutes
Blog
Are you using AI to build training videos? What that means under Article 4 of the EU AI Act
Álvaro Martíez
CEO & Co-founder
Digitization
Are you using AI to build training videos? What that means under Article 4 of the EU AI Act
Most training managers are preparing AI literacy plans for their employees. Few have applied the same logic to their own team — the one that already uses AI tools every day.
If your company uses an AI tool to generate training videos — for onboarding, workplace safety, compliance, or product training — you are deploying an AI system in a professional context. The EU AI Regulation has a name for that: you are an "operator" of that system, and Article 4 already applies to you.
Not to the employees who will watch those videos. To the team members who use the tool to create them.
August 2, 2026 is the deadline for companies to have their AI training plan up and running. For teams that already use AI tools daily, that deadline is immediate.
Article 4 of Regulation (EU) 2024/1689 requires providers and operators of AI systems to ensure that their staff have a sufficient level of AI literacy.¹ The scope is broader than most companies realize on first reading.
"Operator" doesn't mean only the company that built the system. It means any company that deploys an AI system in its activity. If you use a generative AI video platform, an AI writing assistant, predictive analytics tools, or any system that automates decisions or generates content, your company is an operator of those systems.
That includes HR and L&D teams that use AI tools to create training. They are the first employees in the company to interact with AI systems on a professional, ongoing basis. By definition, they are the first people covered by the AI literacy training obligation.
The obligation isn't met with an information session. The Regulation refers to a "sufficient level" of AI literacy, which in practice means employees who use AI systems need to understand:
How the system they use works at a basic level: what type of AI is behind it, what it can and cannot do, and what its known limitations are.
How to verify and supervise the results the AI generates, without assuming the output is always correct or complete.
What the risks are associated with using that system in their specific work context: potential biases, generation errors, privacy implications.
What responsibility they carry as operators of the system toward the people who receive the content they generate.
That training content must be documented, individually accreditable, and available for inspection. It's not enough for employees to "know how" to use the tool. There needs to be evidence that they received structured training on responsible use.
The deadline to have the plan in place is August 2, 2026. That's not the deadline to have completed all the training — it's the deadline to have the plan designed, documented, and in the process of execution.
For companies already using AI tools, that means identifying within the next few weeks which employees use AI systems, what training they've received on their use, and how to certify that training in a formal inspection.
The Regulation's penalty framework is still being developed at member-state level, but non-compliance with Article 4 is a defined offense. The most serious penalties under the Regulation can reach €35 million or 7% of global annual turnover.¹ Having evidence that the plan exists and is being executed before that date is the strongest position in any potential inspection.
If you want a complete step-by-step framework, we've put together a downloadable guide on how to prepare your company for the AI Act: download it here.
One factor makes this an opportunity for training managers: if you already create modules with AI tools, you have the production workflow you need to build your own AI literacy module.
The content to cover is specific and contained: what the system you use is, how it works, its limitations, the criteria for reviewing the output, and the operator's responsibility. That's a 15–20 minute module, not a certification program.
And the distribution and record-keeping infrastructure you need to comply with Article 4 — individual traceability, completion export, accreditable evidence — is exactly the same infrastructure you need for any other compliance module. If you already have that architecture for workplace safety or GDPR, the AI literacy module gets added to the system without building anything new.
The article on how to design the mandatory AI training plan covers the full requirements and how to structure the plan for your entire workforce.
Companies that already use AI to create training sit at a particular intersection: they are both producers of training tools and people who need to receive training. Article 4 applies to them on both counts.
The good news is they also have the technical capability to solve it faster than most. The module format exists, the production workflow exists, the record-keeping infrastructure can be the same.
What's needed is to scope the content, document it, and distribute it before August 2. If you want to work through how to do that with your current setup, get in touch.
It applies to both providers (those who develop the systems) and operators (companies that deploy them in their activity). If you use a generative AI, analytics, or automation tool in a professional context, your company is an operator of that system and Article 4 applies. Developing the tool is not a requirement for having the training obligation.
The deadline establishes when the training plan must be running, not when all workforce training must be complete. The penalty framework is still being developed in EU member states, but non-compliance with Article 4 is a defined offense. Having evidence that the plan exists and is in execution is the primary legal defense in any inspection.
Not necessarily. The plan can be structured in two layers: general AI literacy training for all employees who interact with AI systems, and tool-specific or function-specific modules for profiles with more intensive or higher-risk use. A separate module for every tool isn't required if the general training covers the baseline competencies the Regulation demands.
The evidence standard is the same as for other compliance training obligations: an individualized record of who completed what training, when, and with what result. A completion record generated by an LMS via SCORM or xAPI is the most robust format for a formal inspection. An email without open-tracking or a group session without individual records is typically not sufficient.
@ 2026 Vidext Inc.
Newsletter
Discover all news and updates from Vidext
@ 2026 Vidext Inc.
