AI literacy under Article 4: what a smaller company actually has to do
Updated: 2026-09
Article 4 of the EU AI Act obliges providers and deployers to ensure a sufficient level of AI literacy among their staff. In practice that means knowing which AI runs in the business, naming the roles affected, training them appropriately and documenting attendance.
Most small and mid-sized companies have been using AI for a while without anyone ever deciding to. A language model in marketing, a translation service in sales, an assistant function that quietly appeared inside the accounting software. Nowhere is it written down who knows about it and who understands what the tool does.
That is precisely where Article 4 lands. It requires no course and no mandatory certificate, but a sufficient level of competence — measured against role, deployment context and the people affected by the output. The practical obstacle for smaller companies is rarely the knowledge. It is how to demonstrate, when asked, that the knowledge is there.
How it works
- 01
Make AI use in the business visible
A plain list of every tool with an AI function, including functions that appeared later inside existing software. Without that list the scope of the training duty cannot be determined at all.
- 02
Look at roles rather than headcount
Distinguish between people who select or operate AI systems and those who occasionally use them. The level of competence required differs considerably and justifies different depths of training.
- 03
Choose a training format per role
A structured course with a certificate for roles carrying responsibility, a documented internal briefing for general users. One uniform programme for everyone is either too shallow or needlessly expensive.
- 04
Record attendance so it holds up
Name, date, content and the link to the role. Those four fields decide whether training counts as evidence — not the quality of the slides.
- 05
Define what triggers a refresh
Set out when retraining happens: on new joiners, on material change in AI use, and on new guidance. A date in the calendar is more durable than good intentions.
Why it fits here
Innopulse Academy is aimed at businesses without a compliance department: the course conveys the structure of the AI Act through examples from smaller-company practice rather than corporate structures.
The verifiable certificate solves the part that usually fails internally — the evidence. Anyone who attended holds a record that stands independently of your own filing discipline.
FAQ
Does Article 4 apply to Swiss companies?
Switzerland has not adopted the AI Act. Swiss companies are nonetheless in scope when they place AI systems on the EU market or their output is used in the EU — which happens regularly for software and service providers.
Is one-off training enough forever?
The regulation names no interval, but a record from 2025 says little about today’s system landscape. A refresh makes sense as soon as use changes materially or new guidance appears.
What counts as a sufficient level of competence?
The regulation sets no threshold. The yardstick is the technical knowledge, experience and context of the person concerned, together with the people the system is used on. A documented judgement is worth more than a number.
Do temporary and contract staff need training too?
What matters is the activity, not the employment form. Anyone working with an AI system on the company’s behalf falls within scope. For short assignments a brief documented briefing is usually sufficient.
Innopulse Academy
Innopulse Academy is aimed at businesses without a compliance department: the course conveys the structure of the AI Act through examples from smaller-company practice rather than corporate structures.
The verifiable certificate solves the part that usually fails internally — the evidence. Anyone who attended holds a record that stands independently of your own filing discipline.