I get that this sucks but what's the alternative? It would take legislation that will never pass to eliminate OpenAI's liability for the content that their model spits out and they're already under a microscope by regulators and the public.
This doesn't have anything to do with their liability or regulations, it's what their paying customers want. A company that pays them to provide a chatbot for their business doesn't want the chatbot to say controversial things or tell people to commit suicide.
Actually if it's about customers not paying them if they spit out "kill yourself" content, then I 100% understand that.
These are businesses. Those servers cost money. That compute costs even more. The AI experts, (real AI experts by the way, not tensorflow and pytorch monkeys), cost big money as well. Someone better be paying for all that.
So if the people who want offensive content are willing to pony up the dollars, then great. I've got no problem with giving them what they want. But if they want Barbie or Cocomelon to pay for the offensive content, then yeah, they can be safely ignored. Block as much as you like. (Or rather, block as much as Barbie wants blocked. Which is probably more than you like, but they're paying you handsomely for it.)
Alternative would be to have model itself aware of sensitive topics and and meaningfully engage with questions touching such topics with that awareness.
It might be a while till that is feasible, though. Until then, "content safeguards" will continue to feel like overreaching, artificial stonewalls scattered across otherwise kinda-consistent space.
