AI's mental health fix: stop pretending it's human

If chatbots didn’t assume personas, users might be less vulnerable to them — but many enjoy the illusion.