Among other things, the internet has given rise to a new phenomenon known as Deepfakes. Essentially, Deepfakes are AI-generated content presented visually through images and videos. However, in recent times, Deepfakes have gained notoriety for all the wrong reasons. In this article, we will explore the reasons behind their negative publicity, provide a detailed explanation of what a Deepfake is, delve into how it functions, and offer guidance on identifying them.
Table of Contents
What is deepfake?
Deepfake is a way of manipulating a media, be it an image, video, or audio to represent it as something differently. As the name the name suggests, a deepfake is created or generated using deep and powerful techniques which include machine learning and artificial intelligence. The end result is visual or auditory content that can be decisive in nature.
Taking into account a real instance, in 2021, a deepfake video of the Hollywood actor, Tom Cruise went viral on TikTok in which he was seen attending an award show. A VFX and AI artist actually created the video and many people believed that it was real.
How to identify deepfake content
Here are some ways to identify deepfake content:
- Crooked facial expressions: The easiest way to identify a deepfake content is by looking at the facial movements and gestures of the person in question. If the expressions or movements don’t seem natural and are out of sync, it’s likely the that media is generated using deepfake techniques.
- Inconsistent rendering: Deepfake images or videos of a person such as a prominent personality may carry rough inconsistencies on their face or body which may include improper rendering of hairs, eyes, and even mouth. Basically, if the image or video appears to be artificially altered, there’s a high chance it could be a deepfake.
- Unrecognisable voice or audio: Deepfake videos tend to have uncanny audio which may appear robotic or unnatural. Poorly implemented deepfake audio is often stuttering and sounds broken.
- Know the context: If you’re suspicious about whether the content you’re viewing is deep faked or not, try to understand its context and figure out if the subject in question will actually appear or perform as it’s being depicted in the video or image.
How are deepfakes created?
Deepfakes work by using artificial intelligence (AI) to manipulate images, videos, and audio recordings. Deepfakes can be used to make it appear that someone is saying or doing something that they never actually said or did.
- Deepfakes are created using powerful artificial intelligence systems that can make someone or something act differently, with or without wrong intentions. This involves the use of an AI called generative adversarial network (GAN). The network comprises a generator and a discriminator.
- The generator comes up with media content such as images, videos, and audio recordings, while the discriminator’s task is to differentiate between real and fake content.
- Vast sets of data such as images, videos, and audio recordings of the person for whom the deepfake is to be created are fed into the generator. Once the data is accumulated, it is trained by analysing the data sets and prepares to create a deepfake of the intended person or subject.
- Upon getting successfully trained, the generator begins generating a new image, video, or audio recording as per the instructions. The discriminator on the other hand checks whether the newly made content is real or fake.
- Unless the discriminator gets fooled, the generator will continue to produce the media with the help of analysed data. Once the discriminator judges the deepfake as false media, the process will come to an end.
- This will indicate that the deepfake is successful.
Can deepfakes cause any damage?
Deepfakes often have negative consequences which can lead to tarnishing the reputation or image of someone. It can also play a major role in spreading misinformation over the internet. Here are some ways in which deepfakes can cause damage:
- Blackmailing: Deepfakes can be used to blackmail someone by creating a morphed image or video of them with the intention of releasing it online. For instance, fraudsters may contact individuals to pay them large sums of money after deepfaking their image taken from social media platforms.
- Negative sentiments: With the help of deepfake techniques, an opposing group of people may take a prominent personality’s video and create deceptive content out of it which eventually spreads hatred and misinformation. For instance, a deepfake video could be used to make it appear that a celebrity is speaking wrongfully for a person.
- Put businesses at risk: Deepfakes can ruin the brand image of businesses if an altered or manipulative type of media is surfaced online that directly targets a company’s internal management or operations.
FAQs
Are deepfakes illegal?
The legality of deepfakes depends on their intention and context. Deepfaking a politician to make him appear hateful in front of a large crowd may be considered illegal and unethical while correcting a small detail or effect in movies using the same technology may not be illegal.
Can deepfakes be a threat to digital and information security?
Yes, deepfakes pose a major threat to digital and information security, especially in this day and age where internet resources are easily accessible. Using deepfake techniques, hackers can create a fake avatar of a company’s staff and access crucial and confidential information. This can result in a big loss for the company.
Can deepfake detection tools guarantee 100 percent accuracy?
Deepfake detection tools may not guarantee 100 percent accuracy. These tools may be fooled by the presented image or video. However, if you’re doubtful, it’s still a great practice to run the media through these tools to check its authenticity. Some deepfake detection tools available online include Deepfakedetector.ai and Intel’s FakeCatcher.
