How to Choose an EU AI Act Training Solution
Five solution categories are commonly used to meet the AI literacy obligation: in-house training, a professional association briefing, a general e-learning course, a consulting engagement, and a dedicated AI Act product. None is inherently right or wrong — the choice depends on company size, the risk of the AI you use, and what record remains afterward as evidence.
Article 4 does not prescribe a specific product, a minimum number of hours, or an exam — the Hungarian-language knowledge base confirms this in detail. That cuts both ways: good news, because there is no mandatory procurement list; less good news, because the market is consequently full of conflicting claims. This page does not tell you which vendor to pick — it lays out the axes worth deciding on, including when the answer is not us.
Five solution categories — when each fits, and when it doesn't
These categories describe solution types actually available on the market, not any specific vendor. Where a named type of provider exists — professional associations, international e-learning platforms, consulting firms — only the category's general characteristics are described, never a specific company's price or quality.
1. In-house training built with your own resources
Order of magnitude: no license fee — the real cost is staff hours spent preparing and maintaining the material.
Small headcount, someone on staff who understands the regulation and keeps it current, and simple, well-understood AI use.
Nobody has the time to track legal changes regularly, or the training is not differentiated by role.
What evidence remainsDepends entirely on how systematic the record-keeping is. A bare attendance sheet is weak evidence; a regularly maintained internal log is strong.
2. A periodic briefing from a professional association or chamber
Order of magnitude: often bundled into membership dues, or a nominal participation fee.
The company already belongs to such a body and needs a general, industry-level starting point.
If the company needs ongoing, role-by-role documented compliance — a once-a-year general briefing does not cover that.
What evidence remainsTypically a participation certificate that is not tied to the company's own AI use or a specific role's risk.
3. A course from a general e-learning marketplace
Order of magnitude: typically a modest per-course, per-user fee.
Fast, low-cost way to build general AI literacy, when the goal is awareness rather than documenting legal compliance.
The large majority of courses on international platforms are not EU AI Act-specific and are not built for a particular member state's legal context — they update on the vendor's own schedule, not when the law changes.
What evidence remainsA course-completion certificate, typically not linked to the company's own AI systems or a specific role.
4. A consulting or legal engagement
Order of magnitude: project-based fees, typically starting well above the cost of a packaged training product.
A complex, borderline legal question — for example, a disputed risk classification, or a company acting as both provider and deployer across several systems.
For a typical SME whose question is simply "what do we need to do for AI literacy," this is usually oversized and slow to deliver.
What evidence remainsA thorough, bespoke legal memo or audit — strong evidence, but its cost and timeline are disproportionate to a simple deployer situation.
5. A dedicated EU AI Act compliance product
Order of magnitude: see AI Work Fluency's own pricing below — this is the category we belong to.
Ongoing, role-differentiated compliance is needed, with a recurring trigger (new hires, legal changes), and a retained record matters.
The company itself develops or places high-risk AI systems on the market (a provider obligation), or already runs its own tailored system — see "When we are not the right choice" below.
What evidence remainsPer-role completion logs, an admin report, exportable records — this category is built specifically around demonstrability.
Twelve questions to ask any provider
This list works regardless of provider — including if you end up not choosing AI Work Fluency. Write down all twelve, and ask for answers in writing.
Is the content role-specific?
Why it mattersA data scientist, a manager, and a customer-facing employee carry different risk and need different content — a one-size-fits-all module does not assess that.
Does it leave a retained, exportable record afterward?
Why it mattersIn a supervisory inquiry, the knowledge itself matters less than whether you can later prove who completed what content, and when.
Does it update automatically when the law changes, or does it freeze a one-time snapshot?
Why it mattersThe legal status changes over time — the 2026 Digital Omnibus is a recent example. A one-off course captures the law as it stood at enrollment, not today.
Is it available in your own language, with accurate legal terminology?
Why it mattersA course built for a different legal context can easily mislead a company about its own EU AI Act obligations.
Was it built for deployer obligations or provider obligations?
Why it mattersThe two are not the same: providers carry the regulation's Articles 8–22, deployers the much narrower Article 26. Most SMEs are deployers.
Does it cite a specific, checkable legal provision, or only make general claims?
Why it mattersIf a provider promises "guaranteed compliance" or never cites an article number, that is worth treating with suspicion.
What is the actual time cost per employee?
Why it mattersA half-day workshop and a set of five-minute modules can satisfy the same obligation, yet the time cost differs by orders of magnitude.
Is it recurring, or a one-time event?
Why it mattersNew hires, legal changes, new AI tools — each of these reopens the question. A one-time event does not keep up with any of them.
Who has ongoing access to the record kept as evidence?
Why it mattersIf the record lives only with the provider and cannot be exported, the company gives up access to its own evidence.
What happens to the record if the contract with the provider ends?
Why it mattersDoes exportable data on past completion remain, or does the evidence disappear along with the provider?
Does it impose requirements the law itself does not?
Why it mattersArticle 4 does not require an exam, a certificate, or a fixed number of training hours — if a provider sells these as mandatory, it is worth asking what they are basing that on.
What does onboarding a new employee cost within the system?
Why it mattersTeams grow and turn over — a solution that requires renegotiation or an extra fee for every new hire behaves very differently over time.
When we are not the right choice
AI Work Fluency is built around deployer-level AI literacy training and documentation — not provider-level compliance consulting, and not bespoke legal opinions. That also limits who it fits well:
- If your company itself develops or places a high-risk AI system on the market. Provider obligations (Articles 8–22 of the regulation) — a risk management system, technical documentation, conformity assessment — require a different kind of preparation than a deployer-level AI literacy programme provides. A consulting or legal engagement is the right path here.
- If your organisation has more than 250 employees and its own compliance team. At that scale, bespoke HR-system integration, tailored legal review, and an in-house legal function typically fit the task better than a standardised, SME-scale product.
- If the risk classification of a system is disputed or borderline. Whether a given system qualifies as high-risk in an unclear case calls for a bespoke legal assessment — that is a legal task, not a training one.
- If you already run a working, role-specific system of your own. If your company already documents Article 4 compliance with updating content and exportable records, there is no reason to switch — at most it is worth checking for gaps.
What AI Work Fluency offers, factually: in the dedicated product category, AI Work Fluency consists of a starter package followed by a subscription. The one-time starter package is priced for the Hungarian market — 49,000 HUF for 1–10 employees, 149,000 HUF for 10–50 employees — and includes documented Article 4 AI literacy training plus a compliance record. After onboarding, the Fluency Academy subscription is 14,900 HUF/month flat for 1–10 employees, or 1,190 HUF per employee/month for 11–50 employees, covering ongoing content updates as the law changes, role-specific training tracks, an admin interface, and a compliance report.
If none of the "when we are not the right choice" points above apply to you, and your deployer status is clear, a two-minute check will tell you which obligations apply.
Not sure which obligations apply to you?
Two minutes, no sign-up — see which EU AI Act obligations apply to your company.
Start the AI Act checkFrequently asked questions
Is any specific EU AI Act training product legally required?
No specific product or vendor is mandated. Article 4 expects providers and deployers to take measures to support the AI literacy of their staff and other people acting on their behalf — how that is implemented is left open, and it can be done with in-house resources.
Is an exam or certificate required to comply?
No. The regulation does not prescribe a minimum number of hours, an exam, or an accredited certificate, and no certificate on its own creates a presumption of compliance.
Can a company be fined for choosing a bad training solution?
A breach of Article 4 by itself does not fall under the regulation's closed list of fine categories. What remains a real risk is a market-surveillance measure or an evidentiary disadvantage during the investigation of a different infringement.
What is the real difference between a dedicated AI Act product and a general e-learning course?
A dedicated product typically updates when the law changes, differentiates by role, and produces an exportable record. A general course provides baseline AI awareness, but is not necessarily EU AI Act-specific or built for a particular member state's legal context.
Is it enough to just forward staff the AI tool's user manual?
In many cases this does not meet the expected standard of care as a documented measure. See the Hungarian-language knowledge base for the detailed reasoning.
Is Article 4's wording likely to change again?
The 2026 Digital Omnibus already softened the text once. Further clarification cannot be ruled out, but this page reflects the currently applicable status, dated clearly.
At what company size does a dedicated AI Act product start to make more sense than an internal solution?
It typically pays off where there are multiple roles, recurring onboarding, and demonstrability matters — a one- or two-person company may be well served by a simple internal approach.
What is the difference between provider-level and deployer-level training needs?
Providers fall under the regulation's Articles 8–22, a much broader obligation set. Most Hungarian and EU SMEs are deployers, covered by the narrower Article 26 — provider-level preparation is usually oversized for a purely deployer situation.
Related deep-dive reading (Hungarian-language knowledge base)
These link to our 100-question EU AI Act knowledge base, written and legally verified in Hungarian — useful if you read Hungarian or want the underlying source questions this page draws on.
- Mit ír elő pontosan az EU AI Act 4. cikke az AI-jártasságról?A hatályos 4. cikk szerint a szolgáltatók és az alkalmazók intézkedéseket hoznak a személyzetük és a nevükben eljáró más személyek AI-jártasságának fejlesztése támogatása érdekében. A rendelkezés kifejezetten kimondja, hogy nem kell egyetlen ember meghatározott jártassági szintjét sem garantálni. Kockázati kategóriától függetlenül mindenkire vonatkozik.
- Van-e kötelező AI-képzés, óraszám vagy tanúsítvány az EU AI Act szerint?Nincs. A rendelet nem ír elő óraszámot, kötelező tananyagot, akkreditált képzést, vizsgát, tanúsítványt vagy külső auditot, és nincs olyan tanúsítvány sem, amely megfelelőségi vélelmet keletkeztetne. A Bizottság AI-jártassági Q&A-ja ezt kifejezetten rögzíti. Belső, saját erőforrásból megvalósított felkészítés ugyanúgy alkalmas a teljesítésre.
- Bírságolható-e az AI Act 4. cikkének, az AI-jártassági kötelezettségnek a megsértése?Önmagában nem. A 99. cikk (4) bekezdésének zárt felsorolása nem tartalmazza a 4. cikket, ezért uniós bírságtétel nem tartozik hozzá, és Magyarországon sincs hozzá rendelt forintösszeg. Aki 15 millió eurós fenyegetést kapcsol a 4. cikkhez, a jogszabály szövegével kerül szembe. Ami marad: nem pénzbeli piacfelügyeleti intézkedések és bizonyítási hátrány.
- Hogyan dokumentálható és bizonyítható az AI-jártassági megfelelés egy hatósági vizsgálatnál?Előírt dokumentációs forma nincs, ezért a bizonyítás szabad. A gyakorlatban az bizonyít, ami a mérlegelés folyamatát és annak végrehajtását együtt mutatja, nem a puszta jelenléti ív. Négy réteg együttese ad érdemi választ: AI-rendszerleltár, szerepkör-kockázat leképezés, a végrehajtás bizonyítéka, és a karbantartás nyoma.
- Szolgáltató vagy alkalmazó a cégem az AI Act szerint, és miért számít ez?Szolgáltató az, aki AI-rendszert fejleszt vagy fejlesztet, és azt saját neve vagy védjegye alatt hozza forgalomba. Alkalmazó az, aki saját felügyelete alatt használ egy rendszert szakmai tevékenységben. A puszta használat, bármilyen intenzív, nem tesz szolgáltatóvá. A szolgáltatóra a 8–22. cikk teljes csomagja hárul, az alkalmazóra a jóval szűkebb 26. cikk.
- Érinti-e az EU AI Act a könyvelőirodámat vagy a kis tanácsadó cégemet?Igen, alkalmazói minőségben. Ha a cég szakmai tevékenységben AI-eszközt használ a saját felügyelete alatt, a 3. cikk (4) szerinti alkalmazó. A rendelet nem cégméret szerint válogat. A tisztán személyes, nem szakmai felhasználásra vonatkozó kivétel csak természetes személyre áll, cégre nem. A teher viszont jóval kisebb, mint amit sokan feltételeznek.
A glossary of EU AI Act terms is also in progress at /fogalomtar. The Hungarian version of this guide is at /valasztas/ai-act-kepzes.