Deepfakes are getting good. People complain about these being used for propaganda, while ignoring the giant piles of propaganda around us all, all the time, from every direction to maintain the status quo.
Like, sure, ban or label deepfakes, or whatever. But by that logic, you’d have to ban CNN
Can you confirm these are actually deepfakes ? even with actual expertise in ML I was expecting this kind of near-perfect level, even with a large learning base, not before like two years or so at least.
Using a look-alike would make it simpler, yes, but there’s no easily visible signs of transitions there. Again, as someone that research and work with this stuff even training a good state of the art model on all the Tom Cruise movies combined would still likely produce some artifacts, hence why I’m asking. Maybe that plus his public interviews and the like was enough, but it’s still impressive if it’s a deep fake.
