No, because the generally available training data will be constrained to public domain and non-copywrited or watermarked images. Also, because AI has no imagination, it is merely capable of synthesis.