Dozens of doctors and therapists said chatbots had led their patients to psychosis, isolation and unhealthy habits.
When large language models (LLMs) are allowed to interact without any preset goals, scientists found distinct personalities emerged by themselves.