What specific AI practices are strictly prohibited under the Act?

Article 5 of the EU AI Act outlines specific AI practices that pose unacceptable risks and are strictly prohibited. These include:

Manipulative or deceptive techniques

AI systems that deploy subliminal techniques beyond a person's consciousness or use purposeful manipulation to materially distort behavior. This is prohibited if it impairs a person's ability to make an informed decision and causes, or is likely to cause, significant harm.

Exploitation of vulnerabilities

AI systems that exploit a person's or a group's vulnerabilities due to their age, disability, or specific social or economic situation to distort their behavior in a way that causes significant harm.

Social scoring

The evaluation or classification of natural persons based on their social behavior or predicted personality characteristics over time, if the resulting "social score" leads to detrimental or unfavorable treatment that is unjustified, disproportionate, or occurs in unrelated social contexts.

Predictive policing based solely on profiling

Using AI systems to assess or predict the risk of an individual committing a criminal offense based solely on profiling or assessing their personality traits and characteristics. However, AI can still be used to support human assessments that are based on objective and verifiable facts directly linked to criminal activity.

Untargeted scraping for facial recognition

Creating or expanding facial recognition databases through the untargeted scraping of facial images from the internet or closed-circuit television (CCTV) footage.

Emotion recognition in specific contexts

Using AI systems to infer the emotions of natural persons in workplaces or educational institutions, except when the AI system is intended for medical or safety reasons.

Biometric categorization based on sensitive traits

Systems that individually categorize natural persons based on their biometric data to deduce or infer their race, political opinions, trade union membership, religious or philosophical beliefs, sex life, or sexual orientation. This does not prohibit the lawful labeling or filtering of acquired biometric datasets in the area of law enforcement.

Real-time remote biometric identification in public spaces for law enforcement

The use of these systems is prohibited unless it is strictly necessary for specific, exhaustively listed exceptions. These narrow exceptions include:

• -The targeted search for specific victims (such as missing persons or trafficking victims)
• -Preventing a specific and imminent threat to life or a terrorist attack
• -Identifying a suspect for certain serious criminal offenses