I believe a service like this could succeed if the initial base model wasn't Stable Diffusion and wasn't trained internet scrapes without the copyright permissions.

Their solution basically just amounts of "Ethically sourced Styles" which still has all the red tape that a normal text2image model has because majority of the data is still unapproved for use in an AI model.

Businesses didn't want to get wrapped up in a pesudolegal model that really has no better legality than base SD.