EU AI Act risk triage (simplified)

Problem

Before building an AI feature, StudyBuddy's team runs a first triage of its EU AI Act risk level (simplified — the result must be confirmed by legal experts).

Input: lines feature: tag1,tag2,... until the end of input. Tags describe what the feature does. Classify each feature by the first matching rule:

PROHIBITED if it has emotion-recognition together with education or workplace (and not medical-or-safety), or has social-scoring, or manipulation-harmful.
HIGH if it has education together with any of admission, grading, learning-steering, level-assessment, exam-proctoring; or it has employment-decision, credit-scoring or biometric-identification.
TRANSPARENCY if it has chatbot, generated-content or emotion-recognition.
MINIMAL otherwise.

Print feature: LEVEL - obligation, with obligation:

PROHIBITED → do not deploy in the EU
HIGH → risk management, data governance, documentation, logging, human oversight, accuracy, conformity assessment
TRANSPARENCY → inform users they interact with AI / label AI content
MINIMAL → no specific AI Act obligations

At the end print the counts: Prohibited: a, High: b, Transparency: c, Minimal: d.

Input:

Buddy: education,chatbot,generated-content
AutoGrader: education,grading
EngagementCam: education,emotion-recognition
SpamFilter: moderation

Output:

Buddy: TRANSPARENCY - inform users they interact with AI / label AI content
AutoGrader: HIGH - risk management, data governance, documentation, logging, human oversight, accuracy, conformity assessment
EngagementCam: PROHIBITED - do not deploy in the EU
SpamFilter: MINIMAL - no specific AI Act obligations
Prohibited: 1, High: 1, Transparency: 1, Minimal: 1

Your solution

Write it here or in your IDE, then paste it. Compile and test it yourself before comparing. Your code stays in your browser — it is never sent to or stored on the server.

← S12.3 Did the AI copy licensed code? A similarity scanner S12.5 Ethics case: the AI-writing detector →