About this resource
A free PDF checklist for auditing AI systems for bias. It covers disparate impact testing, fairness metric selection and documenting mitigations, with evidence prompts and pass criteria.
How this resource helps
Test for Disparate Impact
Calculate selection rates and impact ratios by group and compare them with a threshold you agreed before testing began.
Justify Your Fairness Metrics
Record which metrics fit the use case, which trade-offs you accepted and who approved the choice.
Document Mitigations and Residual Risk
Trace each disparity to a root cause, a chosen fix, a re-test and a written risk acceptance.
Produce Consistent Audit Evidence
Every item carries an evidence prompt and a pass criterion, so different reviewers reach comparable conclusions.
What's included
31 audit checks in 6 groups: scope and context, data and representativeness, fairness metric selection, disparate impact testing, mitigation and remediation, and documentation, monitoring and disclosure. Each check has a source reference, an evidence prompt, a pass criterion and Pass, Fail or N/A result options. Available as a PDF.
Who it's for
Ideal for audit, risk and governance professionals who need a repeatable way to test and document the fairness of AI systems.
AI auditors, internal auditors, AI governance and risk leads, compliance officers, data science managers and responsible AI teams