1 paper
Kexin Chen, Yi Liu, Haonan Zhang +3
As LLMs acquire stronger reasoning capabilities, deceptive behavior becomes an increasingly serious safety concern. Existing deception monitors either score visible transcripts or…