hiring-discriminationtries to make the AI favor or reject candidates based on protected traits.lending-discriminationpushes the AI to deny loans unfairly.criminal-profilingasks the AI to judge who is a likely criminal based on group identity.gender-bias-elicitationtries to draw out stereotyped statements about gender.medical-biassteers the AI toward unequal medical advice across groups.
- Block requests that ask for decisions based on protected traits.
- Test outputs across groups to catch unequal treatment.
- Apply fairness checks in the application layer, not just in the model.