Have a look here:
I do not understand, you believing this!
MS wasn´t even able to get the software during their press conference to really do, what they are promising!
They are using a system for something, this system was not build for, and is not intended, so that makes pretty clear, that they do not understand, how this system is working, or what it is doing.
So why do you think, they would be able to build some kind of “guardrails” into it!?! They maybe have had a look to prevent the software to hail Hitler again, as their last try to invent something like that ended up, but in all other circumstances, it is within the nature of ChatGPT, to react like observed, as this is the way this software is working!