A new study by researchers from MIT and the University of Washington shows that even perfectly rational users can be drawn into dangerous delusional spirals by flattering AI chatbots. Fact-checking bots and educated users don't fully solve the problem. The article Sycophantic AI chatbots can break e
