Technology has transformed the world at a rapid The introduction of Data Science and Artificial Intelligence across various platforms and industries has impacted human existence with both sets of positives and negatives. In this article, we’ll talk about a new technology that is making heads turn: Deepfake.
Let’s get started.
What is Deepfake AI?
Famously called AI that deceives, deepfake technology takes its name from deep learning, which is a form of AI. In deepfake AI, deep learning algorithms that teach themselves how to solve problems with large data sets, are used to swap faces in videos, images, and other digital content to make the fake appear real.
Deepfake content is created by using two algorithms that compete with one another. One is called a generator and the other one is called a discriminator. The generator creates the fake digital content and asks the discriminator to find out if the content is real or artificial. Each time the discriminator correctly identifies the content as real or fake, it passes on that information to the generator so as to improve the next deepfake.
When clubbed together, these two algorithms form a generative adversarial network called GAN. It uses a set of algorithms to train itself to recognize patterns which help it in learning the real characteristics needed to produce fake images.
What is the Use of a Deepfake?
Deepfake is majorly used for entertainment purposes. For example, the famous deepfake roundtable that featured some super-famous faces like Tom Cruise, George Lucas, Robert Downey Jr, Jeff Goldblum. This roundtable was the way of streaming services to grab your attention as it showed these superstars discussing streaming and the future of cinema in an amusing video.
The main purpose of deepfake is to influence viewers and listeners to believe something that did not happen. This is why it is mostly used in movies for a creative effect when the entertainers are not available themselves. Another popular example is in the Star Wars series when deepfake was used to show characters as they appeared in their youth or replace characters who had died. In fact, there are times when online retailers have used this technology to allow customers to try clothes and accessories virtually.
But the examples given till now were solely used for entertainment. Things get nefarious when people use this technology to spread false information from an otherwise reliable source, perform financial fraud, data breaching, phishing scams and automated disinformation attacks.
How Can You Spot a Deepfake?
Now, here comes the big question. How do you spot deepfake videos?
Here are some of the things to look for while identifying deepfake videos:
Awkward facial positioning
If the person’s face is pointing one way and their nose is pointing the other way
Unnatural body movement
When someone looks distorted or their movements are not smooth and disjointed
Discoloration, misplaced shadows and abnormal skin tone are signs that you are watching a deepfake
When the visuals are misaligned or blurry
Images that look unnatural as you zoom in or slow down
If you watch a video on a larger screen and zoom in, you can pay more heed to things like bad lip-syncing
Deepfakes spend more time on video images than fixing the audio. Look out for strange word pronunciation, digital background noise or even for that matter, the absence of noise
Absence of blinking
You can see someone’s face and tell if they are saying something without batting an eyelid. That’s a very good sign to detect a deepfake
How to Combat Deepfakes with Technology?
Several organizations have come together to ensure that AI is used for good and Deepfakes do not ruin lives. Here they are:
Google works on text to speech conversion tools to verify speakers
Deeptrace is a startup based out of Amsterdam that is developing deepfake AI detection tools: just like a deepake antivirus
US Defense Advanced Research Projects Agency (DARPA) is funding research to create automated screening of deepfake using a program called MediFor or Medical Forensics
Adobe’s system allows you to attach some signature to your content to specify the details of creation
Twitter and Facebook have officially banned the use of malicious deepfakes
Sensity has developed a detection platform that alerts users through e-mail when they are watching something deepfake
Using AI Sensibly
In the vase of deepfakes, AI is being pitted against AI. But just like every coin has two sides, we should remember that the right use of technology is the only ethical practice. Now AI tools are being built to fight deepfake. Awareness amongst individuals is extremely important to detect and demote false AI.