Human Oversight Procedure

Purpose

This procedure establishes the step-by-step process for implementing, operating, and continuously improving human oversight measures for AI systems in compliance with EU AI Act Article 14. It ensures that natural persons assigned to oversee AI systems have the competence, authority, and means to effectively understand AI capabilities/limitations, interpret outputs correctly, and intervene or stop AI system operations when necessary.

Applies To

• All high-risk AI systems per AI System Classification Standard (STD-AI-001) classification
• All AI systems classified as requiring human oversight per Article 14(3)
• High-risk AI systems referred to in Annex III, point 1(a) — real-time and post remote biometric identification systems (requiring two-person verification per Article 14(5)). Note: per Art 14(5) second paragraph, this requirement does not apply where Union or national law considers its application disproportionate in the context of law enforcement, migration, border control, or asylum.
• All AI systems used in critical decision-making affecting fundamental rights
• All oversight personnel assigned to monitor AI system operations

Does Not Apply To

• Minimal-risk AI systems (unless explicitly required)
• AI systems with no automated decision-making capability
• Third-party AI systems where oversight is contractually assigned to the provider