Back to today’s briefing

Story perspectives

AI Vulnerabilities Exposed: Fake Condition Misleads Chatbots

4/10/2026

31 5 Full Breakdown

Summary

Story summary

Almira Osmanovic Thunström at the University of Gothenburg conducted an experiment that exposed AI vulnerabilities by introducing a fictitious medical condition, bixonimania.
AI chatbots on platforms such as Microsoft Bing and Google’s Gemini presented fraudulent studies as legitimate.
Cureus retracted a paper citing bixonimania after finding it relied on fabricated research, underscoring risks to AI reliability and medical science.

1 of 1

Story summary

Almira Osmanovic Thunström at the University of Gothenburg conducted an experiment that exposed AI vulnerabilities by introducing a fictitious medical condition, bixonimania.
AI chatbots on platforms such as Microsoft Bing and Google’s Gemini presented fraudulent studies as legitimate.
Cureus retracted a paper citing bixonimania after finding it relied on fabricated research, underscoring risks to AI reliability and medical science.