1 paper · 1 filter
Ann Yuan, Asma Ghandeharioun, Carter Blum +6
While existing evaluations of large language models (LLMs) measure deception rates, the underlying conditions that give rise to deceptive behavior are poorly understood. We investi…