Module 4: AI Regulation · BoK IV.C
Prohibited risk and the banned list
Prohibited risk AI is inherently harmful and restricted or banned in many jurisdictions. Six categories recur, including social scoring, manipulation, certain biometric uses (real-time facial recognition in public spaces, untargeted scraping of facial images), emotion recognition in sensitive settings, and predictive policing.
Inherently harmful systems, restricted or banned in many jurisdictions. Exact lists differ, but six categories recur.
- 🧮 Social scoring & discriminatory profiling → the EU bans public-authority social scoring; South Korea bars unjustified discrimination in education, work or essential services.
- 🪤 Manipulation & exploiting vulnerabilities → AI that exploits the vulnerabilities of children, the elderly or people with disabilities.
- 👁️ Certain biometric applications → untargeted scraping of facial images, biometric categorisation on sensitive traits, real-time facial recognition in public spaces (restricted in the EU; disclosure and oversight duties in US states like Illinois BIPA; compliance and labelling rules in China).
- 😶 Emotion recognition in sensitive settings → explicitly listed in the EU Act for workplaces and schools.
- 🚔 Predictive policing & mass surveillance → especially applications with no clear safeguards or accountability.
- ⚠️ Other rights & safety threats → China's GenAI Measures prohibit outputs undermining social stability; US federal guidance warns against deceptive AI impersonation.
Key terms - quick answers
What is “Prohibited risk”?
AI uses banned outright as threats to rights or safety (e.g. social scoring, manipulation, real-time public facial recognition).
What is “Illinois BIPA”?
Biometric Information Privacy Act imposing disclosure and oversight duties on biometric applications.