Regulation (EU) 2024/1689
Find out which parts of the EU AI Act apply to your system
A guided assessment in everyday language, for people who have to work this out without a regulatory team. It takes about fifteen minutes, and it tells you what is still unresolved as clearly as what is settled.
No sign-up. Your answers are saved in this browser only, and you can delete them at any point. This is a preliminary assessment, not legal advice.
The approach
Five things this tool does differently
Plain questions, not legal ones
You are never asked whether your system is high-risk, or which Annex III entry fits. You are asked what the system does and who it affects. The legal mapping happens in versioned rules you can inspect.
Not knowing is a real answer
Choose "I am not sure" and the assessment carries on. The result lists the fact, why it matters, who inside your organisation would usually know, and what document would settle it. An unresolved fact never quietly becomes a no.
Every conclusion shows its working
Each finding names the facts it used, the article and paragraph it came from, and the page of the Official Journal where you can read it. Nothing is asserted without a source.
Alternative outcomes are computed, not guessed
Where a conclusion depends on an open fact, the tool re-runs the rule with that fact settled each way and tells you what the result becomes. That is arithmetic, not an opinion.
Your answers stay in your browser
There is no account, no server, and no analytics. Nothing is transmitted anywhere, and no part of the assessment is sent to an AI model.
Side by side
The same legal test, asked two ways
What the Regulation says
Annex III, point 4(b). AI systems intended to be used to make decisions affecting terms of work-related relationships, the promotion or termination of work-related contractual relationships, to allocate tasks based on individual behaviour or personal traits or characteristics or to monitor and evaluate the performance and behaviour of persons in such relationships.
That sentence is precise, and it is the law. It is also not something a product manager can map their tool onto with any confidence. Asking them to try moves the legal risk onto the person least equipped to carry it.
What EU AI Policy Checker asks instead
What does it do with those people?
Select everything it does, in the plainest terms.
Monitoring and evaluating performance and behaviour at work.
Task allocation based on individual behaviour, personal traits or characteristics.
Opens guidance on who would know and what document would settle it.
Scope
What you get, and what you do not
What the result contains
- Which provisions may apply, and why, with the article and the page number
- Which facts are still unresolved, and who would know each one
- What the result becomes under each way an open fact could go
- The duties that follow, grouped by the role your organisation plays
- Prioritised next actions, evidence to collect, and questions for a vendor
- A reasoning trace for every finding, and a rule set version you can cite
What it will never say
- That your system is compliant, certified, approved or legally safe
- That a practice is definitely prohibited, or definitely not high-risk
- A single overall score, grade, percentage or traffic light
- A conclusion produced by a language model rather than a written rule
- That an unresolved fact can be treated as settled
Start the assessment
There are 83 questions in the set. You will see the ones your situation opens, which is usually between twenty five and forty. You can go back, change any answer, and watch the result change.