GovernanceCore
Trustworthiness

Human oversight

Working definition

Reviewed 30 July 2026

Human oversight is the combination of authority, competence, information, time, and system controls that enables people to supervise AI operation, recognize limitations and automation bias, interpret outputs, intervene, override, or stop the system when needed.

Context

Why it matters

Nominal review can legitimize automated errors. Effective oversight is designed into roles, interfaces, incentives, staffing, and escalation—not added as a final checkbox.

Operating note

What this looks like in practice

  1. 01Define what the overseer must detect, decide, document, and escalate.
  2. 02Give them training, sufficient context, time, and real override or stop authority.
  3. 03Test for automation bias, alert fatigue, rubber-stamping, and unsafe handoffs.

Sources & further research

Primary authority anchors the definition. Research links add conceptual or operational depth. External sources may update independently; always verify legal duties against the current official text.

  1. Official source

    Regulation (EU) 2024/1689, Article 14

    Requires high-risk systems to support effective oversight and specifies overseer capabilities.

    European Union
    2024    Open source ↗
  2. Official source

    Recommendation on the Ethics of Artificial Intelligence

    Places human oversight and determination within a human-rights-centered ethical framework.

    UNESCO
    2021    Open source ↗