AI 101

Human oversight

Also known as: human-in-the-loop (control) · human review

Security

Human oversight is keeping a person accountable for, and in control of, high-stakes AI decisions.

  • A human approves irreversible or costly actions.
  • A human can audit and override.
  • The bridge between autonomy and accountability.

The principle: the higher the stakes, the more a human should be in the loop. Autonomy is a tool; oversight is what keeps it a responsible one.