The constant and unstoppable apology behavior is due to the "alignment". They RLHFed it into that sorry state.

There isn't a fundamental limit there-- you can just not use an OpenAI product, at least until OpenAI convinces governments to outlaw them... :-/