Scientists Invented a Fake Disease Called 'Bixonimania' — AI Systems Started Diagnosing It Within Weeks
A team at Sweden's University of Gothenburg invented a completely fictitious skin condition called "bixonimania" and uploaded fake research papers about it. Within weeks, major AI chatbots began diagnosing it as a real disease. Even more troubling: the fake papers were then cited in peer-reviewed literature.
The Experiment
- Inventor: Almira Osmanovic Thunström, medical researcher at University of Gothenburg
- Fake condition: Bixonimania — a made-up skin condition causing sore, itchy eyes and pinkish eyelids
- Fake author: Lazljiv Izgubljenovic (photo generated with AI)
- Fake papers: Two preprints uploaded to SciProfiles in April-May 2024
- Clues planted: Many hints were deliberately included to signal the papers were fake
Why the Name Matters
"I wanted to be really clear to any physician or medical staff that this is a made-up condition, because no eye condition would be called mania — that's a psychiatric term."
The name was deliberately chosen to be obviously fake to anyone with medical knowledge.
What Happened
- Weeks after upload: Major LLMs began repeating the invented condition as real
- Within months: The fake papers were cited in peer-reviewed literature
- Implication: Some researchers are relying on AI-generated references without reading the underlying papers
The Deeper Problem
This experiment reveals a cascading failure:
- LLMs scrape preprint servers and blogs during training
- Users ask chatbots for medical advice
- Chatbots regurgitate fake information from their training data
- Researchers cite AI-generated references without verification
- The misinformation becomes part of the academic record
Why It Matters
As AI chatbots increasingly serve as medical information gatekeepers, the bixonimania experiment shows how easily fabricated information can enter the AI knowledge pipeline and become "fact" — even to trained researchers. The Common Crawl → LLM → user pipeline has no built-in fact-checking mechanism.