Summary

• Article 5 prohibitions target specific uses rather than AI scoring, biometrics, or predictive systems generally.
• Social scoring is prohibited when prolonged behavioral or personal-attribute evaluation causes unfavorable treatment in unrelated contexts or treatment that is unjustified or disproportionate.
• Untargeted scraping of facial images from the internet or CCTV to create or expand facial-recognition databases is subject to a blanket prohibition, with targeted collection treated differently.
• Individual criminal-risk assessments based solely on profiling or personality traits are prohibited; systems supporting meaningful human judgment based on objective facts generally shift into the high-risk category.
• Emotion-recognition systems are restricted in workplaces and educational institutions, with narrow medical and safety exceptions and unresolved questions around intentions, voice, and behavioral signals.
• Biometric categorization that infers listed sensitive traits is prohibited, while some dataset labeling, filtering, aggregated analysis, and non-sensitive categorization remain outside the ban.
• The practical boundary between prohibited, permitted, and high-risk systems depends on cumulative legal conditions and interaction with the GDPR, the Law Enforcement Directive, and sector-specific rules.