Article 14

Article 14 requires effective human oversight of high-risk AI. The hard part is designing oversight that is genuine rather than nominal, without making the system useless.