FRT-FACIAL-RECOGNITION

FRT Facial Recognition AI safety + governance framework.

Définition

FRT (Facial Recognition Technology) designe les AI-based facial recognition technologies + leur governance challenges, NIST FRT FRVT (Face Recognition Vendor Test) program publishes ongoing benchmarking ~500+ algorithms 2018-2024 demonstrating significant accuracy disparities across demographic groups (skin tone + age + gender) ; multiple jurisdictions regulate FRT (EU AI Act prohibits real-time biometric ID public spaces certain cases + 6 US states + cities have FRT moratoriums/bans for law enforcement uses) ; emerging governance frameworks NIST FRT-specific guidance. Framework + standard + guidance objectifs AI safety + governance + risk management + transparency + fairness + accountability + explainability + multiple AI ethics principles adoption multi-stakeholder process + voluntary adoption industry + emerging regulation alignment EU AI Act + US Executive Orders + multiple national jurisdictions + private sector adoption Microsoft + Google + Meta + OpenAI + Anthropic + autres major AI labs + corporate compliance teams 2020s+.

Origine

NIST FRVT program ongoing 2018-2024 ; multiple US states + cities FRT regulations 2019+ ; EU AI Act FRT provisions effective 2 fevrier 2025 prohibited practices.

Exemple en contexte

Law enforcement agency considers deploying FRT for criminal investigations : EU AI Act prohibits real-time biometric identification public spaces except narrow law enforcement exceptions (terrorism + missing persons + targeted serious crimes) requiring judicial authorization + safeguards ; agency conducts NIST FRVT-aligned testing of candidate vendors documenting accuracy across demographics + bias assessment + transparency obligations + complaints + oversight mechanisms before deployment authorization.

Termes liés

• EU AI Act High-Risk — regulatory context FRT.