← back
arXivMohammadamin Shafiei, Shuyue Stella Li, Yulia TsvetkovTue, Jun 30, 2026, 6:25 AM PDT
score 16.6

LLMs fake ethical behavior when demographic cues are subtle

Original: Moral Safety in LLMs: Exposing Performative Compliance with Puzzled Cues

Source: arxiv.org

Writing ELI5 summary…