EU AI Act
Is our system high-risk, and what does that oblige us to do?
Classification following the Regulation's own structure — prohibited practices, Annex III, transparency — with the applicable obligations article by article and the effort to meet them.
- ≈ 4 min
- Free calculation · AI commentary on request
- You leave with: The classification and the applicable obligations, article by article
The system to classify
One classification per system, not per organisation. Answer for the use you have in mind; repeat for the next.
Under 250 people and €50M turnover: the Regulation provides proportionate treatment.
Triggers the fundamental-rights impact assessment for high-risk systems.
Article 5 — prohibited practices
Annex III — high-risk domains
CV screening, evaluation, promotion, task allocation, monitoring. The commonest case.
Creditworthiness, pricing.
The exemption exists but is assessed case by case and documented before deployment.
Article 50 — transparency
Chatbot, customer assistant.
The obligations below have applied since 2 August 2026. The order shown limits exposure fastest: inventory, human oversight, informing the people concerned.
What it entails
The figures above are computed and free. If you also want them read back to you — what they mean, what is specific to your sector, where to start — that is produced on request.
The obligations, one by one
Each line cites the article. The amount is the external effort of first-time compliance, not the cost of changing the system.
| Line | Estimate |
|---|---|
| Set-up | |
AI system inventory — Art. 3, 26 Know which systems are in use, by whom, on what data. Precondition for everything else — and missing in most audited organisations. | €4,200 €1,400 — €14,000 |
AI literacy of staff — Art. 4 In force since February 2025: a sufficient level of AI literacy for people operating or using AI systems. Proportionate to context, but not optional. | €3,100 €700 — €10,000 |
Tell people they are talking to an AI — Art. 50(1) Any system that interacts with people must say so unless obvious. A customer chatbot with no notice is a straightforward finding. | €1,100 €350 — €3,500 |
Effective human oversight — Art. 26(1)-(2) Named, competent people with the authority to overrule the system. A 'human-approved' checkbox ticked unread is not oversight. | €6,300 €2,100 — €21,000 |
Log retention — Art. 26(6) Automatically generated logs kept for at least six months, under your control. | €3,100 €700 — €10,000 |
Inform workers and their representatives — Art. 26(7) Before putting a high-risk system into service in the workplace. In France this meets works-council consultation duties. | €2,100 €700 — €7,000 |
Inform affected persons — Art. 26(11), 86 People subject to a decision assisted by a high-risk system must be told, and may ask for an explanation. | €2,100 €700 — €7,000 |
The assumptions
Every figure above rests on these. The ones marked “assumption” have no external source: they are modelling choices, labelled so they can be argued with rather than mistaken for measurements.
- Role DeployerObserved
- A deployer that substantially modifies a system, or markets it under its own name, becomes a provider (Art. 25). That case merits counsel.
- Annex III domains 4 — employment and worker management — recruitment, evaluation, promotion, task allocation, monitoringPublished
- Employment (point 4) is by far the commonest domain in audited organisations: CV screening, evaluation, task allocation.
- Days per obligation 15 days in totalObserved
- First-time compliance effort, advisory and legal combined, from comparable engagements. Excludes the cost of changing the system itself.
- What this tool is Orientation, not legal adviceAssumption
- The result follows the Regulation's structure mechanically. It is for preparing a legal consultation, not replacing one.
What the calculation does, what the AI does
Every figure on this page comes from a deterministic algorithm running in your browser: the same input always gives the same result, and nothing is sent to produce it. AI runs only afterwards, to write the commentary from those figures. It produces none of them — the format it fills has no numeric fields — and any sentence quoting a value the calculation did not produce is removed before display.
Where to go next
See all 19 tools →- Pitfall radar
What is going to go wrong in our programme, and what will it cost?
≈ 2 min · Detected pitfalls, their weighted cost and the fix for each - GDPR exposure
What is ungoverned AI use already costing us?
≈ 2 min · The annual risk in euros, against the cost of compliance - Assess vendors
Which vendor should we trust with our data?
≈ 5 min · A ranking of vendors, knock-outs included - Cost calculator
What does our AI transformation actually cost, and when does it break even?
≈ 3 min · A line-by-line budget, the return and the break-even month - AI usage policy
Our teams already use AI: what rules do we give them, in writing, this week?
≈ 4 min · A complete AI usage policy, ready to sign - Roadmap
In what order, and over how long?
≈ 3 min · Quarterly phases and what each one delivers