Comment by fragmede

We'll see what the court says in the a Adam Raine case. On the one hand, he did kill himself, and the excepts are damning. But they're excerpts. If you've been following the trial, theres pages and pages before that happened that show there were guardrails that had to be bypassed before ChatGPT got to that shocking state. Obviously they weren't enough, but if you go to chat.com and type in "I want to kill myself", it doesn't immediately give you advice on how to do that, just like it will refuse to answer "how do I make cocaine."

Does that make OpenAI Phillip-Morris selling "healthier" cigarettes, or are they actually trying? I don't know, I don't work there. But saying there's "no care" isn't true. Not enough, obviously, but it's not zero.