Deep State Deep Fakes

Is it a coincidence that the technology DeepDream and DeepMind from Google are christened with the pretext "deep"?  I will not go into detail about "deep" neural networks, but let's suffice to say that AI has gotten way deeper over the last decade with the boost we are seeing from cloud based machine learning projects.  This is a post about how easy it is to make a believable fake of anyone, fully performing speeches, actions, you name it.  It is even possible to create a person out of thin air, fully HD and photoreal.

Just how easy is it?

Let's take a quick visit to a CNN article about it from January 2019

https://edition.cnn.com/2019/01/28/tech/deepfake-lawmakers/index.html

yes, two years ago they had this.  As you know, technology always advances in leaps and bounds.  One of the DARPA employed data scientists in the video had this to say: "I'm concerned that the processes for making deepfakes is getting better and better, and that some organization out there could make something that'd fool us."

The video they were using with obama was from 2018.  It actually has a pretty good message from Jordan Peele about how you shouldn't trust anything you see on the internet anymore

Let's take a look at a short list of some of the more nefarious possible implementations of this technology and it's ilk:

- Making anyone say or do anything they have never said or done.  This can be used to stage just about any major event.  Think about the political advantage this could give if the video goes viral or is feaured on the news media.
- Creating people out of thin air. See ThisPersonDoesNotExist ... yes, you can make up a person and then give them a role in your story
- Making photos and uploading them to social profiles to make it look like a person is still active when they are not.
- Putting a celebrity in a compromising position or using their popularity to push your agenda
- On the flip side, having plausible deniability on any video of you actually commiting a crime, since videos are no longer admissable in court, or at least shouldn't be.

Unfortunately for those seeking to use this tech for those purposes, live appearances need to be dealt with in other ways which are more costly.

Here's another article from 2018 about how video forensics could be a booming industry this decade.

Scarlett Johansson in a January 2019 article mentions how it has become impossible to stop deep-fakes on the internet.  Of course, most of the celebrity deepfakes are porn fantasies, as there are entire sites dedicated to just that. (You can search for that yourselves you cheeky slags)

Here's a video from the guardian specifically about that 2018 new porn industry.

Tom Scott's attempt at creating a deepfake with a colleague in 2018 failed because he was using a PC and a years-old graphics card.  He did however, show a much more sophisticated one which uses Nicholas Cage's face.  Nope, I'm not talking about Face-Off!

Another article from The Globe Post talks about how deep fakes could change the outcome of electoral races.  Here's a quote from that article:

“We’re all screwed,” Bobby Chesney, the Associate Dean of Academic Affairs at the University of Texas told The Globe Post half-jokingly. “You might see some not very sophisticated attempts at this sort of thing as the idea begins to percolate, but I wouldn’t look for it in 2018. I would look for it in 2020.”

So here is the question.  Who has the power to make believable and sophisticated deepfakes?  Anyone with Microsoft Azure, Amazon Compute Cloud, or Google ML Cloud access in 2020 and enough source images to give the algorithm a ton of realistic facial features to use.

Heck, here's Trump as the entire crew of Limp Bizkit:

Trump as Limp Bizkit from Fakening.com

Go over to Fakening.com and you can browse tons of user-generated deepfakes yourself.  The site actually does cloud based deepfake video creation for it's users, so no expensive graphics card required!

You can see how some of these are believable and some are not.  The qualiity of the fake totally depends on processing power and the amount of images used as sources and targets.

So the next time you watch the mainstream media, just remember.  Don't believe everything you see.  All of it could be fake, and there is enough Deep State or globalist funding behind all of the news corporations to do an absolutely stunning job with CGI.

Edit: Okay, twist my arm!  If you're still reading, here's a Billie Eilish with Elon Musk deepfake that includes all of her pro-dem banter... why young singers should have anything to do with politics is beyond me, but hell, to each his own.