#Deception

LieCraft: A Multi-Agent Framework for Evaluating Deceptive Capabilities in Language Models

Exploring the safety risks of deception in Large Language Models through a new multi-agent framework.