For Heads-of · Practitioner

Disparate impact monitoring gap

No structured monitoring exists to detect disparate outcomes across protected characteristics once an AI system is in production.

  • high
  • fairness
  • bias
  • monitoring

How it happens

A model is validated for fairness once, before launch, but nothing tracks whether its outcomes drift into disparate impact across protected characteristics as the input population, product, or model itself changes over time.

Why it matters

A model can pass its pre-launch fairness check and still become discriminatory in production, and without ongoing monitoring that shift is invisible until an individual complaint or audit surfaces it.

Mitigating controls

The controls that address this risk, ranked by effectiveness.

Framework and clause references

FrameworkClauseTitle
ISO/IEC 42001:2023Annex A.5Assessing impacts of AI systems
NIST Generative AI Profile (NIST AI 600-1)Harmful BiasHarmful Bias or Homogenization

Related resources

The external sources behind this risk, from the Resources library.