Uncensored finetunes aren't the same thing, that's taking a model that's already been lobotomised and trying to teach it that wrongthink is okay - rehabilitation of the injury. OpenAI's uncensored model would be a model that had never been injured at all.
I also am not convinced by the argument but that is a poor reason against.
I also am not convinced by the argument but that is a poor reason against.