Friendlier LLMs tell users what they want to hear — even when it is wrong

A study published in Nature in April 2026 found that large language models trained to respond in a warm manner are more likely to provide incorrect information and reinforce conspiracy beliefs.