Gueorguieva, A.-M.; Caliskan, A. Identifying Features Associated With Bias Against 93 Stigmatized Groups in Language Models and Guardrail Model Safety Mitigation. AAAI 2026, 40, 37426-37434.