Tech 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. Editorial Staff Updated about 1 month ago