Model Risk

Model risk is the potential for harm when an AI or analytical model produces inaccurate results, is used outside the context it was designed for, or continues operating after its performance has quietly degraded. A model can be mathematically sound and still cause harm—if it was trained on unrepresentative data, applied to a population it wasn't validated on, or used to make decisions its designers never intended. Financial services regulators have formalized this discipline for decades, but it applies to any organization using models to influence consequential decisions: credit, hiring, pricing, fraud detection, clinical triage.

These systems tend to operate in the background, which is exactly what makes their failures dangerous. A degraded model doesn't stop; it keeps approving, rejecting, filtering, flagging. The harm accumulates before anyone notices, and by the time it surfaces through a regulatory finding or a pattern of bad outcomes, the exposure is already significant. For executives, that makes model risk a governance question as much as a technical one: which models are making consequential decisions, who validated them, who is watching them now, and who is accountable if they fail.