Boundary drills: prohibited, high-risk, or transparency?

Lesson 4 of 4 in Prohibited AI Practices: The Unacceptable Tier.

The exam question — and the real-world client question — is rarely "recite Article 5." It is: here is a system; which tier? The traps cluster at two boundaries. Prohibited vs high-risk: the same technology family (biometrics, scoring, prediction) appears on both sides, split by elements like real-time, solely, workplace, unrelated context. High-risk vs transparency-only: consequential-sounding systems (chatbots, deepfakes) often carry only disclosure duties, while mundane-sounding ones (CV parsers) carry the full regime.

Drill the boundary until the elements — not the vibes — drive your answer.

Interactive sorting exercise: Sort each system into its EU AI Act tier. Judge by the statutory elements, not by how alarming the system sounds.

Tool: EU AI Act Risk Classifier — Take the boundary skills to the full Risk Classifier: walk real systems through scope, Art 5, and the high-risk routes, and export a reasoned classification.

Interactive checkpoint quiz (1 questions) — open this page in a browser to take it.