Diffusion versions, to start with found in 2014, which insert "noise" to images until finally These are unrecognizable, and after that take out the sound to produce first pictures in response to prompts. Nevertheless you might not hear of Alphabet's AI endeavors from the informa