Swedish researchers tricked ChatGPT, Copilot, and Gemini with fake Bixonimania
Technology
Turns out, even top AI chatbots like ChatGPT, Microsoft Copilot, and Google Gemini can be tricked by fake medical information.
Researchers in Sweden created a totally fake disease called bixonimania and published phony studies about it.
When asked, the chatbots treated the disease as real and even gave advice based on these made-up sources.
Study warns AI spreads falsehoods
The study highlights how easy it is for AI to spread believable but false information.
Lead researcher Almira Osmanovic Thunstrom cautioned against trusting everything AI says without double-checking.
In response, companies like Microsoft and OpenAI say they are working to make their chatbots safer and more reliable, especially when it comes to health advice.