Disparate Impact

A legal and ethical concept where a seemingly neutral AI system or policy disproportionately affects a protected group, even without explicit intent to discriminate.

When an AI rule that seems neutral actually hurts a specific group. For example, an AI that screens resumes by zip code might unintentionally discriminate against certain racial groups who live in those areas.

Even well-intentioned AI systems can produce discriminatory outcomes. Your AI risk assessment process must evaluate for disparate impact to protect the organisation from legal liability and ensure equitable treatment of all stakeholders.