In this Security Magazine article, Brenda Leong explains the importance of red teaming and the need for constant improvement to prepare for future challenges. As artificial intelligence agents learn to draft and propose goals, exist in a persistent state, and demonstrate “initiative”, what matters from a risk perspective is not whether the AI has “intentions”, but that it has these unrecognized capabilities. The article explores the current methods of red teaming as well as what may need to exist in the future.
<< All Publications
