Free self-assessment

EU AI Act Risk Classifier

Start with a real product scenario, then adjust the answers. The tool turns everyday AI use cases into a practical EU AI Act triage report with risk signals, documentation steps, and source links.

0/13 answered

Choose your AI scenario

The fixed questions are the legal risk signals. These scenario buttons translate common product ideas into those signals, prefill likely answers, and keep everything editable.

Pick a scenario to prefill likely answers, or skip this and answer the questionnaire manually. Scenario presets assume the AI system may be offered to EU users; change question 1 if that is not true.
1. Is the system offered in the EU market, used by EU users, operated by an EU entity, or are its outputs used in the EU?
2. Could it involve manipulative AI, exploitation of vulnerable groups, social scoring, banned biometric scraping, restricted emotion recognition, or restricted real-time remote biometric identification?
3. Does it use biometric identification, categorization, or emotion recognition?
4. Is it used for critical infrastructure, safety, or essential operations?
5. Is it used in education, admissions, testing, scoring, or student assessment?
6. Is it used for hiring, worker evaluation, task allocation, or employment decisions?
7. Does it affect access to credit, insurance, housing, public benefits, or essential services?
8. Is it used in healthcare, medical triage, patient support, or clinical workflows?
9. Is it used for law enforcement, migration, border control, courts, or democratic processes?
10. Does a user interact with it as a chatbot, assistant, or automated advisor?
11. Does it generate public text, images, audio, video, code, or recommendations?
12. Are outputs shown to customers, applicants, patients, workers, or the public?
13. Is there meaningful human review before any consequential decision?

Frequently asked questions

Is this legal advice?

No. It is a practical triage tool. Use it to find escalation signals and prepare better questions for a qualified adviser.

Does the tool save answers?

No. The current version runs in the browser and does not send questionnaire answers to a database.

Why does “not sure” matter?

Unclear EU scope, decision impact, or human oversight can change the classification. Unknown answers intentionally trigger review.

Source status

What this classifier can and cannot decide

The questionnaire is a preliminary triage record. It can flag prohibited-practice questions, Article 50 transparency, possible Annex III or Annex I high-risk categories, and the need for specialist review. It cannot determine jurisdiction, resolve an ambiguous legal definition, or certify compliance.

The current implementation dates were checked on September 9, 2026. The 2026 AI Omnibus moved Annex III high-risk obligations to December 2, 2027 and Annex I product-related high-risk obligations to August 2, 2028. Article 50 transparency rules apply from August 2, 2026, subject to the limited transition described in the Commission FAQ.