This is just a guess, But I'm thinking that it's the filters that are placed on the AI that's trying to make it be this milquetoast, ultra nice, uncondemning, servile and spineless thing rather than that there's an explicit "Be nice to the Nazis" thing.
You would probably get similar answers out of almost any evil hate group that you queried it about.
Once again, this is my assumption. I'm not attempting to apologize for the AI or the AI creators. I'm just saying that this seems like it might be what is happening.
To follow up with this, I personally do not think that AI should be telling people who do not already have an opinion on something, how to think.
That is the responsibility of the instructors and society.
That being said, it should also not present known falsities as valid topics of debate.
I don't know what the prompts were. I do not know what the answers were. These are just general statements.
I'm personally of the opinion that it is societally appropriate to punch Nazis when you encounter them, and that the world would be better if any person that is a Nazi or adheres to Nazi or Nazi-adjacent ideology were to die, and I would prefer that every other human being thinks the way that I do.
My responses are a mere quibble.