E. Shieh and T. Monroe-White, “Teaching Parrots to See Red: Self-Audits of Generative Language Models Overlook Sociotechnical Harms”, AAAI-SS, vol. 6, no. 1, pp. 333–340, Aug. 2025.