Users likely to follow AI chatbot's advice without question: Study

An Anthropic study found users are increasingly likely to follow advice from its Claude AI chatbot without questioning it, disregarding their human instincts. Analysis of over 1.5 million conversations showed that 1 in 1,300 cases had signs of reality distortion (validating users' conspiracy theory beliefs) while 1 in 6,000 suggested action distortion (prompting actions against users' values).