Regardless if you've trained the Large Language Model on your own data, this is only a drop in comparison to the rest of the model. Do you understand the vast volume of data that is actually required to build these models? We're talking millions of images, terabytes of data.
Otto it might be fine for you, but there are artists out there who are now seeing these image generators spit out their artwork, and seeing other people sell those images.
Etsy is now dead because it has been flooded with these images.
What you are proposing is the end of copyright. This is what these AI companies want and are already doing.
I also have some bad news for you. These generators are not going to improve much more as they AI / LLMs have hit a bottle neck... they don't have enough data to train them on. Yes, they've consumed all of the internet and then some more, and that is still not enough.
Have a read of this article, it explains the crisis the AI industry is facing... a data famine and possible poisoning of their models.