AI Impact Assessment

Purpose

An AI Impact Assessment evaluates the potential consequences of deploying an AI system — not just whether it works technically, but whether it treats people fairly, respects their rights, protects their data, and can be explained and challenged. It is the governance document that asks 'should we deploy this?' before the question becomes 'how do we fix what went wrong?'

The assessment is not a one-time gate. It is a living document, updated at each stage of the project lifecycle and retained in full — the revision history itself is evidence of responsible governance.

Where It Fits in Your Document Pack

The Impact Assessment informs everything downstream. Your AI Risk Register draws its initial risk inventory from the impacts identified here. Your Human Oversight Plan references the affected parties identified here. Completing the Impact Assessment early means subsequent documents build on a solid foundation rather than starting from scratch.

What This Template Covers

• Assessment identification: stage, status, version, and assessor tracking
• System identification: owner, deployment context, use case, and risk category
• Affected population scope: direct users, non-user populations, vulnerable groups, and scope gaps
• Impact domain assessment across 7 domains with in-scope flag, risk level, findings, and mitigation
• Assessment timing cadence with dated completion fields for each stage
• Framework requirements checklist: NIST AI RMF, EU AI Act Article 27, Canada AIA, ISO 42001, UNESCO
• Findings and escalation: fixable findings, escalation-required findings, and overall recommendation
• Review and sign-off with three-tier approval: assessor, PM/sponsor, governance
• Revision history table and a completion guidance page with field-by-field instructions

Download