Rule vs.Consequence: Dissociable Internal Representations of Moral Reasoning in LLMs

Eugenie Shi

Large language models are increasingly deployed in morally and legally sensitive contexts, yet little is known about how they internally represent distinctions between moral reason…