Enhancing LLM Safety Through a Theoretical Minimax Game Lens

Yihe Deng, Yu Yang, Junkai Zhang +2

The rapid advancement of large language models (LLMs) necessitates effective mechanisms to ensure their responsible deployment by accurately distinguishing unsafe content from beni…