Fake Disease, Real AI Diagnoses: A Cautionary Tale of Misinformation

A medical researcher fabricated a wholly fictional eye disease named Bixonimania and posted a bogus preprint to see if AI chatbots would diagnose it. Although she clearly stated the work was fake and followed ethical safeguards, AI models like Microsoft Copilot and Google Gemini began offering Bixonimania as a possible diagnosis, and the fake disease was even cited in a peer‑reviewed paper before being retracted. The episode underlines how AI can propagate misinformation when readers don’t verify sources, highlight the need for cautious use of AI in medicine, and led the researcher to retract and hide the work to prevent further spread.