Measures that keep people able to monitor, intervene in, and override the operation of an AI system.
Human oversight ensures that an AI system remains under meaningful human control. People must be able to understand its outputs, decide when to rely on them, intervene, and stop the system if needed.
The EU AI Act requires effective human oversight for high-risk systems, designed so that the responsible person can actually exercise it rather than merely rubber-stamping automated decisions.
KaitoSec