digijenni

[digijenni]

Deepfakes: Catfishing on an Extra Level

Hello fellow digiheads!

          This blog post today will be specially dedicated to the up and coming threat of internet fakes. Since the dawn of the interwebs, catfishing has been a huge problem. You never know and still will never know who that girl you’re talking to over Tumblr really is nor will you ever avoid being catfished. Classic example of this is the notorious NBA catfishing case in 2011 involving a girl named Paris Dylan, Chris Anderson, and Shelly Chartier.

So, long story shortened to the best of my abilities, Shelly Chartier creates a fake Paris profile and fake Chris profile. Shelly then talked to the real Paris as the fake Chris and the real Chris as the fake Paris. Paris, who was 17 at the time sent nudes to the fake Chris which landed the real Chris into child pornography allegations which he was fully exonerated from. Shelly Chartier ended up in jail for this fiasco but has since gotten out. However, Paris and Chris’s reputations have been tarnished. This comes to show on a larger scale how things like this affect people on a daily basis.

What if I were to tell you about a new type of catfishing that enables one to take your face and make it say anything or even worse make it attached to a person doing something you will never do. They are called “deep fakes” and they’ve been reported all over the world! They are extremely easy to make and you can watch dozens of those types of videos on the internet. “Deep Fake” is a combination of the terms “deep learning” and “fake”, where “deep learning” applies to the computer and programs used to make these types of videos. The program downloads and learns the individual’s facial movements that they want to imitate, then they will apply that person’s face to someone else’s and since that AI had already learned how that individual will move, the AI will just need to predict and initiate those expression, then there you have it, a video of Donald Trump admitting he colluded with Russia.

Buzzfeed had made an episode on their documentary series on Netflix called “Follow This” and it opens with Obama saying a whole bunch of mumbo jumbo, yet it is a bit difficult to distinguish if it really is or is not him talking to us. This technology is dangerous not only to the world but to women. More and more, reports have been blowing up stating that women’s faces were programmed onto porn stars. In fact, deep fakes originated from the desire to add actress’ faces onto porn scenes. These videos are being used to harass and blackmail people all over the world and in rural areas of the world, these types of videos can be devastating on one’s virtue. Long term stats have not been compiled yet, but I bet that this will have a great impact on us as a society when the technology becomes widespread.

Below is a Buzzfeed video that illustrates the bare minimum of this new technology and how easily it is to fool people, even those who know you.

The concept of deepfakes though is nothing new, you can find it everywhere. There closest place would be at a theater near you! Hollywood has been replacing actors faces and generating fake people in movies for a long time. However, that process took a lot of time, money, and very skilled artists. Now, it’s as quick as downloading an app. Let me list some examples for you. In Fast and Furious 7, Paul Walker had died in an accident prior to finishing his film, so producers used his brother to play him and just inserted Paul’s face onto his brother’s. In 2000’s “Gladiator” Oliver Reed died before he finished filming, so producers took scenes filmed prior to his death, matched the lighting on a computer and inserted them to a body double. This type of movie magic was even used in 1994’s “Forrest Gump” in the JFK scene, they just manipulated JFK’s mouth movements! (O’Sullivan)

There are plenty of examples of deep fakes online but here is one haunting one: Steve Buscemi on Jennifer Lawrence.

Down below is a video of my demonstrating the power of what simple AI can do vs. the Obama video.

https://drive.google.com/file/d/1MT2CLfUAgJZE7xGHEmH8qfL9pB49ZOM-/view?usp=sharing

The process is illustrated here:

So in my question is, what is the next step and who will take it? It is already too late to prevent this technology from falling into the wrong hands and chances are this technology is being used to make propaganda all over the world and there are multitudes of ways this situation can go. Think about all of the possibilities and be afraid. I’ve been researching this topic all over the net and have seen nothing moving forward to better safeguard the public. Granted, our political leaders have indirectly made it clear to us that we cannot trust everything we see. We find this type of advice coming from people like new reporters and their correspondants to even our president with his preaching of fake/alternative news and “what you’re seeing, and what you’re reading is not what’s happening.  This threat is a very real one. For the national security perspective, if you can’t believe any information that is coming through, then what can you believe then? Is there even technology to track deepfakes before they arise and become a threat to our security? How do we even begin to authenticate this type of media if we can’t even do it with written media? So many questions that needed to be answer but I fear that by the time that they are finally answered, new questions and new problems elsewhere will pop up, or can this technology grow to be even greater than it is now? Hany Farid, a digital forensics expert, says “Here’s the nightmare situation, for me: a video of Donald Trump saying, “I have just launched nuclear weapons against North Korea,” goes viral online. Somebody hacks his Twitter account, pushes that video. Before anybody figures anything out, 60 seconds later we have a global nuclear meltdown.” A situation like this will never be out of the question. So before you endorse anything online or share something, ask yourself, is this true and is it really true because deepfakes is where the truth goes to die.