Course contentsShow
← Machine Learning and Deep Learning
Lesson 3501 of 3,53877. Regulation and GovernanceFree lesson

The EU AI Act: Risk-Based Classification

How the EU AI Act categorizes AI systems into unacceptable, high-risk, limited-risk, and minimal-risk tiers.

The EU AI Act: Risk-Based Classification

What you'll learn: How the EU AI Act organizes AI systems into four distinct risk tiers to match regulatory requirements with potential harm.

The Four-Tier Framework

The EU AI Act takes a risk-based approach to AI regulation, similar to how building codes vary based on whether you're constructing a garden shed versus a hospital. Instead of one-size-fits-all rules, systems face stricter requirements as their potential for harm increases.

Unacceptable Risk

These AI systems are completely banned. Examples include social scoring by governments (like China's social credit system), real-time biometric surveillance in public spaces (with narrow exceptions), and manipulative AI that exploits vulnerable groups. Think of this as "forbidden territory"—no development or deployment allowed.

High Risk

These systems can operate but face strict compliance requirements: mandatory conformity assessments, detailed documentation, human oversight, and continuous monitoring. Examples include AI used in hiring decisions, credit scoring, law enforcement, and critical infrastructure. If your model decides who gets a loan or a job, expect rigorous scrutiny.

Limited Risk

These systems have transparency obligations only. Chatbots, emotion recognition systems, and deepfake generators must clearly disclose they're AI-generated or AI-driven. Users deserve to know when they're interacting with a machine.

Minimal Risk

This covers everything else—spam filters, AI-powered video games, recommendation systems for entertainment. No special obligations apply, though general consumer protection laws still hold.

Why This Structure?

The tiered approach balances innovation with protection. Low-stakes applications face minimal friction, while systems affecting fundamental rights face proportionate safeguards.

Key Takeaway: The EU AI Act uses a four-tier risk classification (unacceptable, high, limited, minimal) to apply stricter regulatory requirements where AI systems pose greater potential harm to individuals and society.