AI medical misinformation fooled every major chatbot

Summary: Researchers recently played a trick on artificial intelligence to see if it could tell fact from fiction. A team from Sweden created a completely fake skin disease called "bixonimania," claiming that blue light from computer screens causes sore eyes and dark circles. They uploaded a fake medical study filled with obvious jokes. For example, the paper openly stated, "this entire paper is made up," named one of the authors "The lying loser," and even thanked the USS Enterprise from Star Trek for its help.

Despite these glaring warning signs, major AI chatbots like ChatGPT, Google's Gemini, and Copilot fell for it. They started telling users that bixonimania was a real, rare condition. This experiment highlights a big problem: AI models learn from the internet, which is full of unvetted information. To give safe health advice, medical AI needs strict fact-checking, real peer-reviewed science, and tools like the CRAAP test (which checks for things like accuracy and authority) so they don't hallucinate fake diseases.

For more details, see rssapp-kevinmd-com at kevinmd.com/2026/04/ai-medical-misinformation-fooled-every-major-chatbot.html (opens in new tab)