Human Oversight

The capacity for humans to monitor, evaluate, and intervene in AI system operations to prevent or correct harmful outcomes. A core requirement of the EU AI Act for high-risk AI systems and a principle in all major AI governance frameworks. The appropriate level of oversight depends on the risk level and context of the AI system.