[SOLVED] : Why do we need a solution for Deepfake?
Deepfake has become one of the most questionable phenomena in the tech industry. As this technology introduced a crazy but innovative concept to the market, it has gone more viral than ever.
With many top TV shows and podcasts promoting it, Deepfake has caught the attention of laymen as well. While some are in favor of this technology, many tech experts push back on implementation. But why? What did Deepfake do that made so many anxious and happy at the same time? Read more.
Deepfake: What does this really mean?
Deepfake is a combination of two words, “Deep Learning” and “Fake”. It replaces the person’s face in an image or video with another person that looks like it. Audios, videos and images made using Deepfake are going viral on social media, especially on Facebook, after its Deepfake Detection Challenge.
When users need to know if the person in the video is original or is replaced by someone else, Facebook claims to be running this campaign to undo the effect of Deepfake videos and stop this technology from doing any further harm. However, it is impressive how Deepfake works and covers a person’s face and effectively replaces it with someone else’s.
How does Deepfake work?
Using the concepts of deep learning, this technology effectively replaces faces in images or videos, and guessing the change is nearly impossible due to the accuracy. It puts artificial intelligence and machine learning to work to generate great results.
Using neural network architectures, Deepfake produces extraordinary results that can fool the eyes of the beholder. Neural network architectures such as autoencoders or generative antagonist networks (GANs) aiming to set values are equal and ultimately this produces a great copy that looks the same.
While it sounds like a fun activity, there are some perks that make this technology amazing to beat. However, since the coin has two sides, this technology is held responsible for any unwanted effects that do not suit today’s world.
Benefits of Deepfake
Artificial intelligence has introduced many new technological concepts that are revolutionizing the way things work. However, depending on the context, a particular concept is used, they remain either constructive or destructive. And likewise, Deepfake has the advantage that has made it popular around the world.
The implementation of Deepfake via mobile applications became a reality when a Chinese Zao application implemented it for its customers. Using the app, users can easily swap faces with their favorite celebrities on videos and pictures and post them online. This app went viral in no time and attracted many people to download it immediately.
Moreover, these apps allowed users to create Deepfake videos in minutes which not only
2. True to Body: Body Double
Using Deepfake, only the facial area below the forehead is covered and swapped to create an exact impression. However, a Japanese company managed to make a full head-to-toe swap possible. This particular app not only allows you to swap your face, but gives you full support for body swapping. This helps to provide accurate dual images and videos of the body.
But if we list the downsides, these are the main concerns that many people, including world leaders, are in favor of putting control over the use of this technology for. Due to using Deepfake in a negative context, it has become a puzzle for everyone. And scientists are trying to find solutions to it.
Disadvantages of Deepfake
The downsides of Deepfake involve serious issues that directly affect the company and pose a global threat. Technology developed with the best intentions in mind has turned into an enemy with a simple change of context. But how?
1. Editing content can cause serious problems
Politicians and celebrities are not happy with the results this technology brings. As they are commonly known, and their data is widely available on different servers around the world, it becomes much easier to perfectly swap their faces in pixels. Hence, without AI, it becomes impossible to differentiate yourself and find out whether a Deepfake video is there or not.
For example, using Deepfake videos anyone can manipulate people, and this poses a threat to many superpowered countries. A single Deepfake video swapping a public figure’s face and tricking her down the wrong path can break the whole game of this tech.
2. Deepfaking audio
Not only limited to images and videos, but Deepfake has also entered audio streaming where anyone’s audio can be tempered with slight manipulation. As Deepfake allows expression changes according to users’ wishes, the movement of the face and mouth can be controlled effectively. Software development around the world and in universities are researching this and exploring the steps to effectively remove the confusion. But until then, capturing Deepfake audio is almost impossible, and it can actually cause some serious problems.
3. Increase in fraud
As mentioned earlier, Audio Deepfake poses a significant threat which can allow more threats and frauds. For example, a CEO of a UK based company was tricked during the call when a scammer asked him to transfer a huge amount to a Hungarian account and made the voice of the company representative heard mother of the company. However, the CEO did not do it again as the scammer requested over and over again, but the CEO found that no refund had taken place.
4. Celebrity scams
While Deepfake allows people to place their face in celebrity photos, the reverse is also possible. And that leads to generating disturbing content that can hurt their self-esteem. The same applies to politicians. The majority of these videos are created for fun, but some of them are totally inappropriate and intend to damage their image. Many female politicians suffer from the same issues their Deepfakes are created and go viral which are very difficult to tell apart from the originals.
Due to all these problems caused by Deepfake, many researchers oppose its use and demand to ban it. As the commercial use of Deepfake is no longer under control, the ultimate responsibility lies with us to use this technology with the right context in mind. Practicing what is best not only for us but for the whole of society can lead to positive growth of Deepfake and create a positive image of its use.
The use of Deepfake depends on the context for which the user is using it. Whether they do it with good or bad intention, the technology is going to be changed and a better option will soon be delivered to the market. Therefore, it is ultimately up to us as we choose the context in which we use the latest technology.
All of these inventions are meant to improve our lives. However, without following the ethics and the right context to use them, we all make a big mistake, and the one to blame is technology.
We should learn to use technology in the right way and in our best interests to encourage the efforts of the scientists who invent them for us. With the best of intentions in mind, Deepfake can provide great support for many things and prove beneficial in many ways.
Deepfake, a technology as fascinating as it is terrifying
Right now, the hunt for Fake News is very much in fashion. It must be said, with social networks, everyone has the possibility to relay anything. We are more and more inclined to be exposed to this type of “information”. And in the different techniques used to make fakes credible, we find a new technology as fascinating as it is frightening: The Deepfake.
What is this, Deepfake?
Even if it is more and more famous, we might as well give a little summary for the few who do not yet know what Deepfake consists of. The latter, from the name of a Redditer who discovered the process, is a technology using Artificial Intelligence which consists of transposing someone’s face onto the body of another person. Obviously, as one can imagine, the technique was quickly used for pornographic purposes. Indeed, at the beginning of the year, many videos were available on the PornHub site where we could see the faces of celebrities transposed on actors / actresses X. We can also find the technology in uses for humorous purposes, for example the actor Nicolas Cage, recurring target of humorous diversions, was able to see his face transposed in many sequences. Below, a small best-of of its best integrations.
But where it gets more touchy is that these video montages can be used in politics. Nowadays, politicians are highly publicized personalities. You can find many videos of a political figure, during speeches, interviews, etc. A politician’s gallery of expressions can therefore be very comprehensive which allows the AI to create expressions that are much more believable. Here is an example of a video created by BuzzFeed using the Deepfake, it features the former President of the United States making abusive remarks against Donald Trump. A sequence that serves to show the capabilities of the technology, and raise awareness on the subject.
Easy to use
If only this technology was accessible only to people experienced in the field but not, there is an application extremely easy to access: FakeApp, which allows, in just a few hours of tutorial, to create your own Deepfake.
“Yeah, ok, we can put someone’s face on another, but hey, I don’t really see the problem, because video is one thing, but audio is one other ”And I’m fine with that, but now it’s extremely easy to get someone to say things without actually saying it. This was also what Adobe offered with its software Voco, presented in 2016 (but no more news since), which allowed with less than 30 minutes of voice recording of a person to create any sentence with the voice of the recorded person. The two technologies combined have a pretty destructive potential if they are put in the wrong hands, and let it be known, it is highly possible if the use is so easy to access.
Knowing that we already have a hell of a problem with Fake News lately, there is no risk that Deepfake will work out. It is still important to keep in mind that if these technologies evolve, the techniques which consist in countering them will evolve, too, but not necessarily at the same speed.
The Darpa (Defense Advanced Research Projects Agency) has also started to work on the subject since this summer, specialists in “forensics” and digital will meet for a competition. They will have to make videos, images and sounds modified using AI. Each team will try to identify the anomalies in the projects proposed by the opposing team.
One thing is certain, we have not finished hearing about Deepfake.