.png){kind=logo}
In recent years, deepfake has become a term that almost everyone who follows technology has come across at least once. For those more involved in the topic and researching it in more detail, they come across many different related concepts. Cheapfake is one of them!
Also known as shallowfake, cheapfake is essentially a less sophisticated version of deepfake. However, reducing it to such a simple summary would underestimate the differences between these two concepts, which we cannot afford to do when dealing with digital fraud. This is why I want to discuss the differences between the two and the threats they pose in identity fraud.
Deepfake refers to content manipulated using artificial intelligence (AI), usually videos or audio files.
To read a more detailed blog post about deepfake:
Deepfake uses machine learning techniques and swaps the face or voice of a target person with someone else, creating highly realistic but entirely fake media. Suppose the resemblance between the target and source person is high, and the creator is skilled in using sophisticated methods. In that case, the resulting content can be nearly flawless and indistinguishable to the naked eye.
To examine confusing deepfake examples from recent years:
Initially used for entertainment, deepfakes are now employed in crimes ranging from social manipulation and defamation to financial fraud and system attacks. Moreover, access to tools for creating deepfake content has become significantly easier in the past few years.
To see the most common deepfake frauds:
As mentioned earlier, cheapfake (or shallowfake) is content altered using simple video, photo, or audio editing tools instead of advanced AI. Filters used on social media sites are a perfect example of cheapfake. Even without looking closely, you can easily tell that someone using a dog filter isn’t actually a dog, or that the person you're video chatting with isn’t really in the North Pole despite what their background might suggest.
Besides these simple uses, cheapfakes are also often used to create content that causes at least mild discomfort within society. For example, in this image, former U.S. President Donald Trump is said to be striking a victory pose after a shooting, inspired by Adolf Hitler.
Another video example of cheapfake is here:
We see a doomsday scenario created by adding footage from Ankara, Turkey, and screams in the background to a digitally generated storm video. When this video was uploaded to YouTube, it was believed to be real and shared by many online news sites.
After discussing both concepts, we can summarize their main differences as follows:
The rise of cheapfake and deepfake technologies is an increasing threat to identity verification systems. The digitalization of businesses and financial institutions necessitates the remote verification of customer identities, which in turn increases the potential for fraud.
So, what can be done to prevent this? Giving up on the benefits of the digital world and returning to physical verification methods is neither possible nor effective anymore. The first thing we need to do is to analyze the problem correctly. The first question we should ask is: at which stage of identity verification are these methods used? The answer is crystal clear: the biometric verification stage.
A fraudster attempting to perform a transaction with a stolen ID card produces a video using the ID cardholder's photo and tries to pass a liveness test with this content. If security measures are weak, even less sophisticated cheapfake content can easily pass these tests.
To delve deeper into the role of deepfakes in identity fraud:
Access to deepfake tools is now easier and more affordable, increasing the number of potential fraudsters. Therefore, companies need to take more professional measures. Some of these measures are as follows:
AI Detection Tools: As the quality of deepfakes improves, it becomes increasingly difficult to detect these contents with the human eye. Thus, we need to turn to systems trained with AI to detect threats created by AI.
Advanced Liveness Detection: Using advanced facial recognition technologies is crucial. Instead of being limited to the facial expressions of the person in front of the camera, techniques like detecting screen usage or background compatibility should be included in the AI model.
Multi-Factor Authentication: The most crucial step in remote identity verification is always to build a multi-factor authentication model. No matter how good your liveness detection method is, relying on just one method is always risky. Building a robust process that combines ID verification, OTP confirmation, and biometric verification modules is the most reliable way to prevent identity fraud.
Techsign always works hard to stay vigilant against current methods used in identity fraud. Therefore, it has focused its biggest investment on deepfake detection in recent years. It also has advanced products for multi-factor authentication. If you want to protect your company against these growing threats and take action before it’s too late, contact us now!
info@techsign.com.tr
