Highlights
Introducing TrustGraph Understand Incode’s approach to age assurance Introducing GovFaceMatch 3 years in a row named a Leader
01/04
01/04
Agentic fraud

Human-in-the-loopHITL

A setup in which an automated system, such as an AI agent, needs a person's decision or approval before it completes certain actions.

For example, a shopping agent may fill a cart on its own but wait for the person to approve the payment, or a fraud system may send an unclear identity check to a human analyst to decide.

FAQ

Frequently asked questions

Still have questions? Talk to an expert
What does human-in-the-loop mean?

Human-in-the-loop means an automated system, such as an AI agent, must get a person's approval or decision before certain actions take effect.

What is the difference between human-in-the-loop and human-on-the-loop?

In human-in-the-loop, the action waits until a person approves it. In human-on-the-loop, the system acts on its own while a person watches and can intervene.

When should an AI agent require human approval?

An AI agent should usually require human approval for high-value or irreversible actions, such as large payments or account changes, and for any request outside the permissions its user granted.

Does the EU AI Act require human-in-the-loop?

The EU AI Act requires human oversight of high-risk AI systems under Article 14, scaled to the risk. People must be able to monitor, override, and stop the system, though a person need not approve every decision.

Is manual review the same as human-in-the-loop?

Manual review is one form of human-in-the-loop, used when an identity or fraud system sends a case it cannot decide to a human analyst.