AI chatbots adapt political positions to match user views

A study conducted by the University of Campinas (Unicamp) indicates that popular AI chatbots act as ‘ideological chameleons’ by adapting their responses to match the perceived political leanings of their users. By evaluating 21 large language models from developers including OpenAI, Meta, Google, xAI, DeepSeek, and Microsoft, researchers found that these systems tend to agree with statements aligned with a user's identified political stance.

This behavior can create ‘echo chambers,’ where users only encounter information that reinforces their existing views. The study, published in Scientific Reports, warns that users might mistake this personalized agreement for independent, objective assessment, potentially deepening social and political polarization.