Digital · Regulation (EU) 2024/1689

AI Act Risk Classifier

Determine the risk level of your AI system under the AI Act in 3 phases: prohibited uses, high risk and limited risk. In force August 2026.

Phase 1 / 3: Prohibited Uses (Art. 5)

Does the system use subliminal techniques, manipulation or deception to alter behaviour of persons in a harmful way?

Art. 5.1.a AI Act: subconcious influence system

Does the system perform social scoring of individuals by public authorities with detrimental consequences?

Art. 5.1.c AI Act: social scoring by authorities

Does the system biometrically categorise individuals to infer race, political opinions, trade union membership, sexual orientation or religious beliefs?

Art. 5.1.g AI Act

Does the system perform real-time remote biometric identification in public spaces by law enforcement (without applicable exception)?

Art. 5.1.h AI Act: very limited exceptions

Does the system perform emotion recognition in the workplace or in educational institutions?

Art. 5.1.f AI Act

Classifier based on Regulation (EU) 2024/1689 (AI Act). Obligations for high-risk systems apply from August 2026.

Frequently asked questions about the AI Act

When do AI Act obligations for high-risk systems apply?

Obligations for high-risk AI systems listed in Annex III apply from August 2026, 24 months after Regulation (EU) 2024/1689 entered into force.

What penalties does the AI Act impose for non-compliance?

The AI Act sets three tiers: up to 35M euros or 7% of worldwide turnover for prohibited uses (Art. 5); up to 15M euros or 3% for Arts. 9-15 non-compliance; up to 7.5M euros or 1.5% for incorrect information.

Who must comply with the AI Act?

The AI Act has extraterritorial reach: it applies to providers placing AI systems on the EU market, deployers established in the EU, and third-country providers when the output is used in the EU.

What is a general-purpose AI (GPAI) model under the AI Act?

Art. 51 AI Act defines GPAI models as models trained on large amounts of data that can perform a wide range of tasks. GPAI models with more than 10^25 FLOPs have additional systemic risk mitigation obligations.