Jaime Ramirez, CEO of Preventor
Through deep learning, attackers put at their service an army of machine learning algorithms that are capable of spoofing a person's image, sound or video in real time. Remote-controlled avatars to commit identity fraud are one of the biggest threats of this decade.
The growing use of artificial intelligence and its application in digital environments is challenging the perception of human beings, to the point that it is practically indistinguishable if an image or a video has been altered by means of these technologies to communicate multiple messages and with different purposes, as we see from Preventor, where we respond to this with liveness detection.
The multiple applications that have gone viral in recent years in which users take a selfie to have fun with aging filters or to change gender, are simple examples of deepfakes. The World Intellectual Property Organization (WIPO) describes these tactics as the "superimposition of human characteristics on the body of another person, and/or the manipulation of sounds, to generate a realistic human experience".
Deepfakes are practically already part of the Internet landscape, but unless a rigorous fact-checking exercise is carried out or such creations are intended to be publicly disclosed, they are difficult to trace. At Preventor we see this as creating serious challenges in terms of the risks related to the proliferation of disinformation and above all in terms of threats to people's security on the Internet.
Its application has spread rapidly in industries such as film and video games, at the same time that it has been applied for marketing purposes. But it is when its use has a political disinformation background or when it seeks to exploit a security vulnerability that its use becomes problematic. That is when identity verification and especially liveness detection come in to correct these vulnerabilities and close doors to crime.
While this is happening, artificial intelligence continues to develop and is introduced in different daily activities, as we have seen in recent months with the global boom of ChatGPT. Such is the expectation for the growth of this industry that projections by Statista indicate that the market value of this business could exceed the US$300 billion barrier by 2025 worldwide.
Turning our backs on artificial intelligence at a time when it is taking up more and more space in the economy and society is surely a bad decision. The most viable and effective thing to do is to respond to this opportunity with the necessary technological preparation and tools to be able to ride this new wave smoothly.
At Preventor, we increasingly perceive a trend to adopt liveness detection solutions, which force to take identity verification to another level, allowing organizations to detect if behind a screen there is really a live person or not.
Let's remember that the area of authentication is now one of the most sensitive in the technology industry and that errors related to this discipline generated million-dollar lawsuits to companies. In 2019, for example, the technology giant Apple was sued by a New York student for having linked him to a series of robberies in the company's stores based on the results of a facial recognition system used by the company.
Authentication systems are vital for companies that are performing online onboarding processes or continuous verification, so liveness detection is an increasingly adopted tool in all platforms that require login such as banks or fintechs, which are also highly exposed to deepfakes.
Basically, the system detects movements and still images that try to mimic a real person and circumvent the security of different platforms. The technology shields companies from deepfakes because it is able to recognize methods such as the use of filters, avatars or even masks, saving them a lot of headaches.
The liveness detection infrastructure includes a series of constantly learning algorithms that are connected to deep artificial intelligence neural networks.
The most advanced liveness detection systems currently include innovations such as texture detection and artificial intelligence (AI), which can accurately differentiate between the skin of a human being and a recaptured image, whether video or photograph.
This type of tools, whose infrastructure is supported by biometrics and artificial intelligence algorithms, can differentiate between 3D and 2D fraud attempts, as well as remotely controlled avatars that are capable of falsifying not only a moving image of a person but also his or her voice.
This type of tools, whose infrastructure is supported by biometrics and artificial intelligence algorithms, can differentiate between 3D and 2D fraud attempts, as well as remotely controlled avatars that are capable of falsifying not only a moving image of a person but also his or her voice.
The value of these solutions lies in the fact that they are able to capture different biometric traits simultaneously (such as eye or facial recognition), thus creating additional layers of security that will be more difficult for malicious actors to penetrate and prevent them from implementing replay techniques.
Given that this occurs in real time in situations critical to the security of any company or user, such as joining a service or authentication in an application, there are also mechanisms to increase security such as response challenges. Through these, users must interact with the system and validate themselves after completing random steps that involve the use of biometric technologies.
With this scenario, at Preventor we urge more organizations to explore more about the potential of liveness detection to respond to the challenges generated by artificial intelligence in this revolution that it is already representing in different productive activities and also in daily life. Therefore, it is clear that the best way is to integrate technologies in time and prepare teams for all the disruptions that lie ahead. The best response to these challenges is innovation and companies still have room to do so, because in the very near future the integration of artificial intelligence will be the difference between the survival or the end of an organization in an automated world.
