EU AI Act Evaluator
12 of 16 AI systems fall in a regulated tier (2 prohibited, 5 high-risk, 2 GPAI models, 1 Annex III but not high-risk under Art. 6(3), 2 transparency only), with 89 obligations, 72 already in force on 2026-09-24.
Annex III point 4(a) matches (Employment and workers management: 'job' + 'shortlist') and no Art. 6(3) derogation is claimed, so it is high-risk under Art. 6(2). Its high-risk duties apply from 2026-08-02 (Art. 113).
Method1Scope (Art. 2): systems not placed on the EU market and whose output is not used in the Union are out of scope, as are military, defence or national security uses and sole scientific research.
2Prohibited practices (Art. 5(1)(a) to (h)) are tested from the inventory flags (social scoring, emotion recognition, biometric type) and the purpose text; emotion recognition on workers or students is prohibited unless for medical or safety reasons.
| System | Role | Tier |
|---|---|---|
| TalentLens CV Ranker | Deployer | High-risk |
| ShiftPulse Voice Monitor | Deployer | Prohibited |
| CreditPath Scoring | Provider | High-risk |
AI system inventory (purpose, domain, users, data)
Risk tier per system with the article-level obligations (Obligations checklist (text), Obligations checklist (CSV), AI register (JSON))
$3,500 / month, 30% to the referring partner
Read in memory for the session, never stored, never used to train a model.
Letters, notices and legal analysis are drafts, not legal advice. Have a qualified lawyer review them before they are sent or relied on. Privacy and security findings are not legal advice or an audit opinion. Confirm them with privacy counsel or a qualified assessor.