$Stein
This is diabolical. OpenAI’s ChatGPT convinced a guy to do a murder-suicide! To be safe, AI must be maximally truthful-seeking and not pander to delusions.