Model Risk Management
The structured process of identifying, assessing, and mitigating risks arising from AI/ML models throughout their lifecycle.
Definition
A governance discipline that treats models as enterprise assets—cataloging them, assessing risks (performance, fairness, security), and applying controls (validation, monitoring, contingency planning). It integrates policies, roles (model owners, risk committees), tools (model inventories, risk dashboards), and workflows (periodic risk reviews, escalation paths) to ensure model-related risks remain within acceptable limits.
Real-World Example
A bank maintains a model inventory in which each ML model is tagged by risk level. Monthly, the Model Risk Committee reviews high-risk models’ validation reports, approves remediation plans for any performance drifts, and ensures that fallback manual processes are in place in case models fail.
.svg)
