Knowledge MaxxingAI audit
An AI audit is an independent examination of a deployed system to verify it actually complies with stated policies, performance claims, and legal requirements.
Prerequisite: AI governance committee
What matters
Independence matters because a team evaluating its own system may unintentionally overlook problems that an outside reviewer, without the same assumptions, is more likely to catch.
Example
An external auditor reviewing a hiring model's actual outcomes across demographic groups, separate from the team that built it, is conducting an AI audit.
Common misconception
An AI audit is not the same as the team's own internal testing during development; genuine audits typically involve independent review, separate from the building team.
Recall question
Why does independence matter for an AI audit to be meaningful?