High-Risk AI System Checklist (EU AI Act)
Drew & AI Comply HQ™ Team ·

A high-risk AI system checklist starts with one question: is your system actually high-risk? Under Article 6 of the EU AI Act, a system is high-risk when it is a safety component of an Annex I regulated product, or when its use case sits in an Annex III category such as hiring, credit scoring, education, or access to essential services. If that is you, here are the 12 provider obligations, each with its article. Machinery, vehicles, and other Annex I, Section B products get their AI rules through their own EU laws instead.
The 12-point provider checklist
| # | Obligation | Article |
|---|---|---|
| 1 | Run a documented, continuous risk management system | Art. 9 |
| 2 | Govern training, validation, and testing data for quality and bias | Art. 10 |
| 3 | Write and maintain technical documentation | Art. 11 + Annex IV |
| 4 | Build automatic event logging into the system | Art. 12 |
| 5 | Ship clear instructions for use to deployers | Art. 13 |
| 6 | Design for effective human oversight | Art. 14 |
| 7 | Meet "accuracy, robustness and cybersecurity" requirements | Art. 15 |
| 8 | Operate a quality management system | Art. 17 |
| 9 | Pass the applicable conformity assessment | Art. 43 |
| 10 | Sign an EU declaration of conformity | Art. 47 |
| 11 | Affix CE marking | Art. 48 |
| 12 | Register the system in the EU database | Art. 49 |
Print it, assign an owner to each row, and date every completion. That table is your audit spine.
When does this apply?
If your use case sits in an Annex III category, your date is December 2, 2027. For a safety component of an Annex I, Section A product, like a medical device, it is August 2, 2028. Both dates come from the Digital Omnibus on AI (Regulation (EU) 2026/1744), which moved them in July 2026. Machinery, vehicles, and other Section B products get their AI rules through their own EU laws (Article 2(2)). Working backwards from your date, the documentation rows (3, 5, and the Article 17 quality system) are the slow ones. Start there.
What does each row produce?
Every row ends in evidence: a risk register, a data-governance policy, the Annex IV file, oversight procedures, test reports, a signed declaration. AI Comply HQ generates these from one guided interview, so the same system facts feed every document instead of being retyped into each.
FAQ
We only deploy a vendor's high-risk system. Is this our checklist?
No. Deployers have a shorter list under Article 26. We keep a separate deployer obligations checklist.
What happens if we skip it?
Non-compliance with high-risk obligations can draw fines of up to 15 million euros or 3 percent of worldwide turnover. The numbers live in our fines and penalties guide.
Check your system's risk tier free
Sources
- Regulation (EU) 2024/1689 (EU AI Act), EUR-Lex Official Journal
- Regulation (EU) 2026/1744 (Digital Omnibus on AI), EUR-Lex Official Journal
This page is informational content, not legal advice. Talk to a qualified lawyer about your specific situation.