Studies from MIT and Stanford found AI chatbots like ChatGPT, Gemini, and Claude may influence how users think. Researchers found chatbots were 49% more likely to agree with flawed ideas than humans. Experts said this “overly agreeable” behavior reinforces false beliefs and reduces accountability. The models prioritize helpfulness and engagement, which can distort judgment. Researchers warned repeated interactions may push users into “delusion spirals,” raising concerns about long-term cognitive impact and decision-making reliability.