Glossary

Human Oversight

Human oversight is the set of technical and organisational measures that let people understand, monitor and intervene in what an AI system does, including the power to override or stop it, so that responsibility for its decisions stays with a person.

Term: Human Oversight

Human oversight is the set of technical and organisational measures that let people understand, monitor and intervene in what an AI system does, including the power to override or stop it, so that responsibility for its decisions stays with a person.

Why it matters

Article 14 of the EU AI Act requires providers to design high-risk AI systems that people can oversee effectively while in use. Article 26 requires deployers to assign that oversight to people with the necessary competence, training, authority and support. Both apply only to high-risk systems, from 2 December 2027 (Annex III) and 2 August 2028 (Annex I).

UK GDPR Article 22A treats significant decisions about individuals as solely automated where there is no meaningful human involvement in taking them. A reviewer prone to automation bias, who approves outputs without challenge, may not meet that test.

How it works in practice

Our AI Security Programmes help clients match an oversight model to each decision by what is at stake: human in the loop, human on the loop or human in command. Clients name each overseer; we help record their training and authority to intervene. Our FAQ covers which oversight model fits which AI decision.

Related terms: human agency, automated decision-making, AI governance and algorithmic accountability.

Want this in context?

See how this term fits into the broader programme of work.