Some users are already experiencing problems with AI chatbots, including self-radicalization through conversation spirals with overly sympathetic systems that reinforce rather than challenge user beliefs.