Think Before You Lie: How Reasoning Improves Honesty

ArXi:2603.09957v1 Announce Type: new While existing evaluations of large language models (LLMs) measure deception rates, the underlying conditions that give rise to deceptive behavior are poorly understood. We investigate this question using a novel dataset of realistic moral trade-offs where honesty incurs variable costs. Contrary to humans, who tend to become less honest given time to deliberate (Capraro, 2017; Capraro, 2019), we find that reasoning consistently increases honesty across scales and for several LLM families.