What Is AI Risk Tiers?
AI Risk Tiers is the EU AI Act's risk-based classification of AI systems into unacceptable, high, limited, and minimal risk, each with different obligations.
AI Risk Tiers — the EU AI Act's risk-based classification of AI systems into unacceptable, high, limited, and minimal risk, each with different obligations.
Rather than regulating all AI equally, the EU AI Act sorts systems by risk. Unacceptable-risk practices (such as social scoring) are banned; high-risk systems (Annex III uses like employment, credit, and biometrics) carry the heaviest obligations; limited-risk systems (such as chatbots) mainly owe transparency duties; and minimal-risk systems are largely unregulated. Knowing which tier a system falls into is the first governance question for any EU-exposed deployment.
Source: EU AI Act (risk-based approach)
Plain-language explanation
Rather than regulating all AI equally, the EU AI Act sorts systems by risk. Unacceptable-risk practices (such as social scoring) are banned; high-risk systems (Annex III uses like employment, credit, and biometrics) carry the heaviest obligations; limited-risk systems (such as chatbots) mainly owe transparency duties; and minimal-risk systems are largely unregulated. Knowing which tier a system falls into is the first governance question for any EU-exposed deployment.
Related terms
See where you stand on AI governance
Take the free 7-question maturity assessment and get a personalised action plan.
Free assessment — 3 minutes →