The article mentions that the main reason for suspecting that results were fabricated is that many images from the researcher's papers seem to have been manipul

The article mentions that the main reason for suspecting that results were fabricated is that many images from the researcher's papers seem to have been manipulated. As someone who is not particularly familiar with AI but occasionally picks up on it here on HN, I am now wondering how difficult or easy it is these days to take a manipulated image of a western blot and feed it to an AI tool trained on unmanipulated western blots to create a deepfake that is then virtually indistinguishable from a true one. -- Can perhaps anyone here familiar with deepfake technology provide an assessment?