Could flattering AI make humanity turn on itself?

A study shows chatbots mirror users' politics, part of a broader tendency known as AI sycophancy. Researchers fear this could deepen polarization.