Large language models exhibit bias towards Western entities in non-English, non-Western languages, impacting cultural adaptation and fairness.