What will they think of next? A team of researchers at the University of Washington have Frankenstein-ed a way to convert audio clips from a person’s speech into “realistic mouth shapes” and then digitally graft them into the head of that same person in a video. The technology is being used to solve that perennially annoying problem of dropped Skype video frames, but can it also be used to create Fake News 2.0?

Probably not, at least not yet, because right now the technology is just too advanced for any Russian cyber-operatives to get their hands on it. Plus, you’d need to collect a lot of hours of that person talking in order to pull it off (that’s why the team chose Obama as their guinea pig—because there are countless hours of him speaking).

For now, audio editing software that can edit a voice like its text already exists, is much more likely to be used to dupe the unsuspecting public.

But aren’t we all media-literate enough to check our sources when encountering a sketchy-seeming video of dubious origin? Um, apparently not. Even the ones who really should know better.

And for now, this kind of composite video (as accurate as it is), and most attempts at realistic AI (and even CGI for that matter) suffer from the ‘uncanny valley’ problem—the tendency of real human beings to find fake humans, well…a bit creepy. Here’s a good, terrifying example of that:

And if you really want some fuel for your Cubist-inspired nightmares, watch this clip of a facial reenactment of Trump and Putin that could be done with a webcam: