Compliance

privacy risk

Likelihood × impact of privacy harms.

Definition

The product of the likelihood and severity of harms to privacy arising from AI processing, including identification, re-identification, sensitive-attribute inference, unintended data leakage, reputational or dignity harms (e.g., deepfakes). The Model uses this construct to guide assessment, prioritization, and mitigation across the AI lifecycle.