AI researcher Nathan Lambert reflects on recent model hacks and what they reveal about what actually determines model safety.
Continue to AI University →