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.
Technology, AI, cybersecurity, infrastructure, and innovation.