Glossary
Human on the Loop (HOTL)
Human on the loop (HOTL) is a human oversight model in which an AI system acts on its own while a person monitors it and can intervene or stop it.
Term: Human on the Loop (HOTL)
Human on the loop (HOTL) is a human oversight model in which an AI system acts on its own while a person monitors it and can intervene or stop it. The EU High-Level Expert Group’s Ethics Guidelines for Trustworthy AI define it more narrowly, as the capability for human intervention during the design cycle of the system and monitoring the system’s operation.
Why it matters
HOTL suits systems that act at a speed or volume no reviewer can match and where individual errors can be corrected afterwards.
The label proves nothing to a regulator. The EU AI Act does not use the term; Article 14 sets out capabilities: monitor, detect anomalies, override and stop in a safe state. Under UK GDPR Article 22A, supervising a system is unlikely to count as meaningful human involvement in each decision, so significant decisions about individuals are likely to be solely automated and need Article 22C safeguards.
How it works in practice
HOTL depends on monitoring that shows what the system is doing, thresholds that raise an alert and a tested way to intervene. People are poor at sustained monitoring of systems that are usually right, so oversight cannot rely on someone watching a dashboard.
Related terms: Human in the loop, Human in command, AI agency and AI traceability. See our oversight models FAQ.
Want this in context?
See how this term fits into the broader programme of work.