Skip to main content

Prohibited AI Practices Under Article 5

Complete guide to prohibited AI practices under EU AI Act Article 5, including social scoring, manipulative AI, and biometric categorization.

Article 5 bans specific AI practices deemed to pose an unacceptable risk. These prohibitions have been enforceable since February 2, 2025.

Violations carry the highest fines: up to 35 million EUR or 7% of global annual turnover.

Enforceable since February 2, 2025

Article 5 prohibited practices are already enforceable. Non-compliance exposes organizations to the maximum penalty tier.

Prohibited AI Practices

Subliminal & Manipulative AI

AI systems deploying subliminal or manipulative techniques beyond a person's consciousness, causing or likely to cause significant harm.

Exploitation of Vulnerabilities

AI systems exploiting vulnerabilities due to age, disability, or social/economic situation to materially distort behavior.

Social Scoring

AI systems by public authorities for evaluating persons based on social behavior, leading to detrimental treatment unrelated to the data context.

Predictive Policing (Individual)

AI systems predicting criminal offence risk based solely on profiling or personality traits.

Untargeted Facial Recognition Scraping

AI systems creating or expanding facial recognition databases through untargeted scraping from the internet or CCTV.

Real-Time Remote Biometric Identification

AI systems for real-time remote biometric identification in public spaces for law enforcement, except in narrowly defined circumstances.

Penalties for Violations

Up to 35M EUR or 7% of global annual turnover

The highest fine tier under the EU AI Act. Proportionate caps may apply for SMEs and startups.

Narrow Exceptions

Very narrow exceptions exist, primarily for law enforcement under strict judicial safeguards.

Law Enforcement Biometric Exception

Real-time biometric identification permitted for searching for victims, preventing terrorist threats, or locating suspects of serious offences. Prior judicial authorization required.

Post-Remote Biometric Identification

Post-remote biometric identification for serious criminal investigation with prior judicial authorization. Classified as high-risk, not prohibited.

Frequently Asked Questions

Related Guides


How Scanara Helps

Scanara automates EU AI Act compliance from code to dossier. Connect your GitHub repos and get compliance reports in minutes.