Computers are becoming increasingly capable of simulating life. Modern cinema, for example, depends heavily on computer-generated sets, scenery, and characters in spite of the practical settings and props that were once traditional, and these scenes are mostly indistinguishable from reality.
Deepfake technology has recently gotten a lot of attention. Deepfakes, the most recent version of computer imagery, was generated when artificial intelligence (AI) was designed to replace one person’s likeness in captured video with another. As an example, I have also put a deepfake video of Mark Zuckerberg in the mid-section of the blog to help you better understand the ramification it could have in our society.
What is “Deepfake”?
The word “deepfake” is derived from the underlying technology “deep learning”, which is a form of artificial intelligence. Deep learning algorithms, which use vast amounts of data to teach themselves how to solve problems, are used to swap faces in video and digital content to create realistic-looking fake media.
How are Deepfakes created?
Deepfakes can be made in a variety of ways. The first is that you run thousands of face shots of the two people into an encoder, which is an AI algorithm. The encoder looks for and learns comparisons between the two faces, then reduces them to their shared characteristics, compressing the images. The faces are then recovered from the compressed images using a second AI algorithm called a decoder. You train one decoder to recover the first person’s face and another decoder to recover the second person’s face because the faces are different. You basically feed encoded images into the “wrong” decoder to execute the face swap. A compressed image of person A’s face, for example, is fed into a decoder that has been educated on person B. After that, the decoder reconstructs person B’s face using the gestures and orientation of face A. This must be repeated on each and every frame for a credible video.
A Generative Adversarial Network, or GAN, is another method for creating deepfakes. Two artificial intelligence algorithms are pitted against each other in a GAN. The generator is the first algorithm, which takes random noise and transforms it into an image. This synthetic image is then applied to a pool of actual images – perhaps of celebrities – that are fed into the discriminator, the second algorithm. The synthetic images would first seem to be nothing like faces. However, if the procedure is repeated several times with performance feedback, the discriminator and generator will both develop. After a sufficient number of loops and reviews, the generator can begin to produce incredibly realistic faces of entirely fictional celebrities.
Negative use of Deepfake
While the potential to turn faces dynamically in order to create convincing and realistic-looking synthetic video has some intriguing beneficial applications (such as in cinema and gaming), it is obviously a dangerous technology with some alarming applications. One of the first real-world uses of deepfakes was to create synthetic pornography.
- In 2017, a Reddit user known as “deepfakes” developed a pornographic website of face-swapped actors. Since then, pornography (particularly revenge pornography) has consistently made headlines, wreaking havoc on the reputations of celebrities and public figures. Pornography accounted for 96 percent of deepfake videos discovered online in 2019, according to a Deeptrace survey.
- Deepfake footage has also been used in politics. In 2018, for example, a Belgian political group, posted a video of Donald Trump delivering a speech in which he called for Belgium to leave the Paris climate agreement. Trump, on the other hand, never gave the speech; it was a ruse. That wasn’t the first time a deepfake was used to produce deceptive images, and tech-savvy political analysts are anticipating a fresh era of fake news featuring convincingly authentic deepfakes in the future.
- Several deepfake videos have recently gone viral, providing millions of people around the world their first glimpse of this new technology: President Obama swearing at President Trump, and Mark Zuckerberg revealing that Facebook’s real purpose is to deceive and manipulate its users.
- The amount of deepfake material on the internet is rapidly increasing. According to a survey from startup Deeptrace, there were 7,964 deepfake videos online at the start of 2019; nine months later, that number had risen to 14,678. It has, without a doubt, continued to expand since then.
It doesn’t take much thought to see the damage that could be done if whole communities are shown fake videos that they mistake for real. Imagine deepfake video of a senator committing fraud or sexual harassment only before an election, or US troops committing crimes against people abroad, or President Trump announcing the deployment of nuclear missiles against North Korea. In a future where there seems to be some uncertainty about whether those clips are genuine, the effects may be disastrous.
Positive use of Deepfake
Many sectors, including movies, educational media and visual communications, gaming and entertainment, social media and healthcare, material science, and various business areas, such as fashion and e-commerce, will benefit from deepfake technology.
Deepfake technologies can help the film industry in a variety of ways. It may be used to create digital voices for performers who have lost theirs due to illness, or to update film footage rather than reshoot it. Moviemakers may be able to reconstruct classic movie scenes or produce new films featuring stars who have passed away.
- Deepfakes’ platform allows for more telepresence and natural-sounding online games and immersive chat environments. This aids in the development of stronger human relationships and online engagement.
- In the same way, innovations can be beneficial in the social and medical sectors. Deepfakes will help people cope with the death of a loved one by digitally bringing a dead friend “back to life”, allowing a mourning loved one to express their final goodbyes.
- Businesses are interested in the future of brand-applicable deepfake technology because it has the potential to drastically change e-commerce and ads. The ability to easily try on clothes digitally is an obvious possible use; the technology not only encourages people to make digital copies of themselves and have these personal avatars travel with them through e-stores.
How to detect a Deepfake
While deepfakes become more popular, civilization as a whole may have to adapt to noticing them in the same manner that internet users have adapted to spotting other types of fake news.
There are a handful of indicators that give away deepfakes:
- Since current deepfakes have difficulty animating faces accurately, the effect is video in which the subject never blinks, or blinks way too often or unnaturally.
- Look for skin or hair conditions, as well as faces that seem to be blurrier than the setting in which they’re placed. The focus will seem unnaturally soft.
- The audio does not seem to fit the user, particularly if the video was faked but the actual audio was not.
- Badly rendered jewelry and teeth, as well as odd lighting effects like erratic illumination and iris reflections, may be a giveaway.
Combatting Deepfakes with technology
Although deepfakes can only become more real as techniques advance, we aren’t completely defenseless in the face of them. A number of firms, including some startups, are creating methods for detecting deepfakes.
- Sensity, for example, has created a deepfake identification tool that works like an antivirus and sends users an email while they’re watching something with AI-generated synthetic media fingerprints.
- Operation Minerva, which detects deepfakes in a more direct manner. The algorithm used by this organization compares possible deepfakes to previously “digitally fingerprinted” footage.
- Facebook also sponsored the “Deepfake Detection Challenge” last year, a free, interactive effort to promote the development of new tools for identifying deepfakes and other types of distorted media. The competition offered cash prizes worth up to $500,000.
Many analysts agree that as technology advances, deepfakes will become much more advanced, posing more serious risks to the public, such as voter manipulation, political tension, and additional illegal activity. As a result, we must be more mindful and cautious in our use of social media sites. Even, we should avoid sharing so much information about our personal lives on our various social media sites. I will leave it to you, the reader, to judge whether the deepfake technology is a blessing or curse to our civilization!
For more technology related information, click here.
Aqib Adnan Shafin
Content Writing Intern