Glossary
AI Impact Assessment
An AI impact assessment, also called an AI system impact assessment or AIIA, is a documented assessment of the consequences an AI system may have for individuals, groups of individuals and society across its life cycle, covering its deployment, intended use and foreseeable misuse.
Term: AI Impact Assessment
An AI impact assessment, also called an AI system impact assessment or AIIA, is a documented assessment of the consequences an AI system may have for individuals, groups of individuals and society across its life cycle, covering its deployment, intended use and foreseeable misuse.
Why it matters
ISO/IEC 42001 requires a process for this assessment in clause 6.1.4. Clause 8.4 covers carrying it out and Annex A.5 supplies the supporting controls. ISO/IEC 42005 gives optional guidance on carrying one out.
The AI risk assessment is aligned with the organisation’s AI policy and objectives, and its results drive risk treatment. The impact assessment examines a given system’s consequences for people and society, and its documented results feed that risk assessment. An auditor will expect evidence of both.
How it works in practice
The organisation sets the circumstances that trigger an assessment and the level of detail, and can group lower-impact systems. Clause 8.4 calls for reassessment at planned intervals or when significant changes are proposed. We advise reusing evidence from an existing DPIA, though a DPIA covers personal data processing only.
Related terms: Fundamental Rights Impact Assessment and Equality Impact Assessment. Compare them in Which Assessments Does an AI System Need?
Want this in context?
See how this term fits into the broader programme of work.