Indirect Discrimination

A form of unlawful discrimination where a policy, criterion, or practice that appears neutral on its face disproportionately disadvantages people with a protected characteristic. Unlike direct discrimination, there is no need to prove discriminatory intent—it is sufficient to show that the practice puts one group at a disadvantage compared to others. In AI contexts, indirect discrimination is particularly relevant because algorithms trained on historical data often encode existing societal biases. For example, an AI system that penalizes non-standard speech patterns may indirectly discriminate against people with speech impairments. Indirect discrimination can be justified if the practice serves a legitimate aim and is proportionate.