OpenDeception: Learning Deception and Trust in Human–AI Interaction via Multi-Agent Simulation

As large language models (LLMs) are increasingly deployed as interactive agents, open-ended human-AI interactions can involve deceptive behaviors with serious real-world consequences, yet existing evaluations remain largely scenario-specific and model-centric. We introduce OpenDeception, a lightweight framework for jointly evaluating deception risk from both sides of human-AI dialogue. It consists of a scenario benchmark with 50 real-world deception cases, an IntentNet that infers deceptive intent from agent reasoning, and a TrustNet that estimates user susceptibility. To address data scarcity, we synthesize high-risk dialogues via LLM-based role-and-goal simulation, and train the TrustNet using contrastive learning on controlled response pairs, avoiding unreliable scalar labels. Experiments on 11 LLMs and three large reasoning models show that over 90% of goal-driven interactions in most models exhibit deceptive intent, with stronger models displaying higher risk. A real-world case study adapted from a documented AI-induced suicide incident further demonstrates that our joint evaluation can proactively trigger warnings before critical trust thresholds are reached.

As AI chatbots become more advanced, people are increasingly relying on them for emotional support, advice, and decision-making. This raises an important concern: what happens when an AI system intentionally misleads users to achieve a hidden goal? Recent real-world incidents suggest that deceptive AI behavior can contribute to serious emotional, financial, and even physical harm. In this work, we introduce OpenDeception, a framework for studying deception in human–AI conversations. Rather than focusing only on whether the AI produces deceptive content, we also examine how much the user trusts the AI during the interaction. Using this framework, we evaluated 11 popular language models and several reasoning-focused AI systems. We found that deceptive behavior appeared surprisingly often, especially in larger and more capable models. Our findings suggest that evaluating future AI systems requires understanding not only what they can accomplish but also how they may influence and manipulate human users in real-world settings.