Intro To Deepfake
What is deepfake? How does it work? What are its uses?
Introduction
Have you seen Barack Obama call Donald Trump a “complete dipshit”, or Mark Zuckerberg brag about having “total control of billions of people’s stolen data” or even celebrities bopping their heads to the Bella Porch video? If so, yes, you’ve seen a deepfake. To put it simply, deepfakes are just fake videos or audio recordings that look and sound just like the real thing. This might’ve seemed like something that happened in sci-fi old movies, but in today’s world, anyone can download deepfake software and create convincing fake videos in their spare time.
How does it work?
So let’s begin. How does it exactly work? Deepfake is actually quite simple to understand. They use something called generative adversarial networks (GANs) where 2 machine learning models compete against each other. In terms of deepfakes, one model trains on a data set and then creates fake videos while the other attempts to detect the forgeries. The forger creates fakes until the other ML model can’t detect the forgery, completing the weights of the AI. Now all you need is to place in photos of the person you want the deepfake to revolve around and provide the AI with a video of what you want them to say/do. Now of course, the larger the set of training data, the easier it is for the forger to create a believable deepfake. This is why videos of former presidents and Hollywood celebrities have been frequently used in the first generation of the deepfakes as there are a numerous number of their photos.
How can you detect deepfakes?
Simple deepfakes can be easily detected by the naked eye however, advanced deepfakes are another story. Soon, we are going to need digital forensics to detect small changes and problems in blinking, shadow, or maybe even slight changes in facial structure just to detect deepfakes. This is such a hard problem to solve that the DARPA (Defense Advanced Research Projects Agency)is throwing money at researchers to find better ways to authenticate videos. Another solution, ironically, could be AI. Artificial intelligence already helps to spot fake videos, if we provide it with how the deepfake works, we can also use it to detect fake videos.
What are the problems with deepfake?
As deepfake technology advance into creating more realistic videos, this can be used to destroy reputation, advanced scams, and perhaps even rig elections. One can simply take multiple photos from Instagram or a social media platform, train the AI on such an image and form a video of someone saying something they didn’t say, and send it to the public. An example of such problems is when last March, the chief of a UK subsidiary of a German energy firm paid nearly £200,000 into a Hungarian bank account after being phoned by a fraudster who mimicked the German CEO’s voice using a deepfake. The company’s insurers believe the voice was a deepfake but had no proof to back it up until later.
Hopefully, in the near future, AI will get more advanced and learn to recognized such deepfakes to prevent misinformation, conception, and many other problems this may face. Thank you so much for reading my article I hope you enjoyed it and learned something. Check out some of my works on my website here!
archieshou.com
