Adversarial Model Degradation Rate (AMDR)

GMI-DEF-003Defense OperationsInfrastructure-DependentPattern O-2: Model Drift / Currency

Definition

The percentage decrease in AI model performance when operating against adversarial conditions compared to baseline performance in routine conditions, measured through periodic red-team testing.

Formula

AMDR = ((P_baseline - P_adversarial) / P_baseline) x 100

Proposed formula, pending validation pilots.

At a Glance

Category
AI Calibration & Trust Integrity

Decision Supported

How vulnerable is this model to deliberate adversary manipulation?

← Training Adversary Diversity Index (TADI)Alert Investigation Yield (AIY) →

Assess Your Organization

The AI-GPS Assessment evaluates your AI governance maturity and recommends KPIs like this one for your implementation roadmap.

Take the Free AI-GPS Assessment