A new study led by Washington State University highlights the importance of tone in how artificial intelligence (AI) corrects medical misinformation. Published in the International Journal of Human-Computer Interaction, the research found that the effectiveness of an AI's correction depends more on the tone it uses than on whether the correction comes from a human or an AI. For people who see AI as a purely technical tool, corrections in a neutral, factual tone are most persuasive. However, for those who believe AI can be more humanlike, an empathetic tone—such as showing understanding or concern—works better. The study involved 857 parents of children in the age range recommended for the human papillomavirus (HPV) vaccine. Participants were shown a simulated Facebook comment thread that started with a false claim: "HPV vaccines increase the risk of neurological problems." An AI corrections account then responded to this claim. The neutral responses used straightforward, direct language, while the empathetic tone was more conversational, such as saying, "I hear you, but scientific studies have shown..." The study found that the correction was most effective at reducing misunderstandings when the tone matched a participant's level of anthropomorphism—the tendency to give human-like qualities to non-human entities. This means that if someone believes AI can be empathetic or understanding, an AI using a warm, empathetic tone would be more convincing. The findings suggest that social media platforms, government agencies, and other organizations could improve AI fact-checking by tailoring the tone of their responses based on users' beliefs about AI. This could be done through a simple onboarding process that assesses users' views on AI and adjusts the AI's communication style accordingly. This study adds to the growing understanding of how to effectively correct misinformation. Previous research has had mixed results on whether an empathetic tone is helpful in reducing false beliefs. This study introduces a new perspective by testing how tone effectiveness varies based on the beliefs and expectations of the person receiving the correction. The research team included Porismita Borah, a professor at Washington State University's Edward R. Murrow College of Communications, along with co-authors Ziyao Zhang, Xiaohui Cao, and Danielle Ka Lai Lee.