Blinded by the Bot: Why AI Chatbots Cue Human Unethical Behavior in Negotiations

Artificial intelligence (AI) is increasingly used in negotiation, with chatbots serving as autonomous bargaining agents. Yet it remains unclear whether people uphold ethical standards when negotiating with these systems. If they do not, understanding the psychological processes that weaken moral judgment and identifying effective mitigation strategies is essential. The current research explores these questions by investigating how ethical fading (i.e., a psychological process in which individuals fail to recognize the moral aspects of a situation) shapes (un)ethical behavior in negotiations with AI agents. Across five pre-registered experiments, we find that ethical fading is more pronounced when individuals negotiate with AI chatbots compared to human counterparts. The first four studies show that this diminished moral awareness increases unethical behavior, specifically selfishness and misrepresentation. The fifth study demonstrates that informing people about the cooperative (vs. competitive) design of the AI chatbot significantly reduces these behaviors. Together, the findings reveal a novel ethical concern in human-AI interaction, showing that the presence of AI chatbots can alter human cognition in ways that make moral lapses more likely. We encourage organizations deploying AI bargaining agents, along with developers, to recognize the risk of ethical fading and adopt human-centered safeguards that promote ethical conduct.