Maybe you shouldn't shut down OpenAI over this. But each instance of a particular ChatGPT model is the same as all the others. This is like a company that has a magical superhuman therapist that can see a million patients a day. If they're found to be encouraging suicide, then they need to be stopped from providing therapy. The fact that this is the company's only source of revenue might mean that the company has to shut down over this, but that's just a consequence of putting all your eggs in one basket.