Last updated:
A formal process to evaluate the potential consequences of an AI system's deployment on individuals, groups, and society at large.
Required under ISO 42001 Clause 6.1.4, this assessment differs from a standard risk assessment by focusing on external harms rather than internal organizational risks. It forces teams to consider dimensions like civil liberties, environmental impact, and economic displacement, mapping how an AI system might change the life of an end-user or community.
Real world example:
An autonomous vehicle company conducts an AI System Impact Assessment to determine how their navigation model affects traffic patterns in low-income neighborhoods versus high-income areas.
