I think the main issue with transformer image generation in this respect is that not only can the image be explicit but also using it for this has an incredibly low effort cost and could photo-realistically depict a real living person and materially affect their life.
Whereas drawing applications have a natural barrier to achieving all of these together: time and skill.