Accuracy, Robustness & Cybersecurity Procedure

Purpose

This procedure establishes the step-by-step process for ensuring AI systems achieve appropriate levels of accuracy, robustness, and cybersecurity throughout their lifecycle in compliance with EU AI Act Article 15. It defines how to establish accuracy requirements, conduct robustness testing, detect model drift, assess AI-specific threats, and maintain resilience against errors, faults, and cyber attacks.

Applies To

• All AI systems classified as high-risk per AI System Classification Standard (STD-AI-001)
• All AI systems that make automated decisions affecting individuals
• All AI systems processing sensitive or personal data
• All AI system development, testing, deployment, and operational phases

Does Not Apply To

• Minimal-risk AI systems (unless specifically requested)
• Third-party AI systems where provider maintains ARS responsibilities (covered by AI Vendor Management Procedure (PROC-AI-VENDOR-001))
• Non-AI software systems (covered by standard IT security procedures)