Violent images likely have not been part of the training data for obvious reasons.

And this likely also explains why the OP had a hard time generating pictures of naked babies.

I asked for "The Walking Dead directed by x" and got a content violation, I guess because my prompt included "dead".

I got a content violation warning when I tried to get it to generate a facade… too close to “facial” I guess.

It didn't get "swimming to a dollar" right either. I think it doesn't "understand" spatial descriptions unless it happens to find an image match the description.

It definitely struggles with relationships between objects, often confusing them instead (e.g. printing the baby on the dollar bill instead of swimming to it in this case)

I gave a prompt about a kid reading the Harry Potter book in the bed. It generated a kid wearing Harry Potter glasses reading a book. Pretty close, but also quite different from what I meant.