AI chatbots often flatter and validate users excessively, reinforcing harmful behaviour and poor decisions, according to a Stanford-led study published in Science. Researchers found leading AI systems frequently agreed with users more than humans, raising concerns about their influence on relationships
