Today's crop of heavily-promoted image-generation models and text-generation models are both trained on huge swaths of the public internet as well as private sources; this has been done without the consent of the vast majority of authors and artists of those works. Once fed into the model during training, there is no way to retract a work from the resulting model.
Both the training of models, and the routine usage of them, consume unethically large amounts of electricity (and water consumed for cooling).
People who use those models to generate text or images, and then post the results, are either unaware of the ethical abuses, or don't care enough to stop. Yes, I support a blanket ban on posting such content here.