Risk levels

Prohibited AI practices

Art. 5

Article 5 of the regulation establishes a list of AI practices strictly prohibited due to their nature contrary to Union values. Prohibited practices include social scoring systems, subliminal manipulation, exploitation of vulnerabilities, real-time remote biometric identification in public spaces (with exceptions), and biometric categorisation based on sensitive characteristics.

Related terms

Biometric identification system

An AI system intended to identify natural persons by comparing their biometric data to a reference database.

Biometric categorisation system

An AI system intended to assign natural persons to specific categories based on their biometric data.

High-risk AI system

An AI system subject to the full set of obligations under the regulation due to its area of application.

Related obligations

Consult the regulatory obligations associated with prohibited ai practices in our article guide.

See all AI Act obligations

Is your AI system affected?

Free diagnostic in under 3 minutes. Identify your risk level and obligations under Regulation (EU) 2024/1689.

Start free diagnostic