What is a Deepfake?
Deepfakes are synthetic or fake media files used to alter existing imagery, video, or audio. Deepfakes usually include a specific individual manipulated and replaced with another person’s face or voice. This work is done using generative artificial intelligence (AI)-powered neural networks, also known as Generative Adversarial Networks (GANs), that process information, create patterns, and learn much as the human brain does.
Today, the widespread availability of sophisticated computing technology and the growing accessibility of AI enables virtually anyone to create highly realistic fake content. The number of Deepfake videos available online is increasing by 900 percent annually. The manipulation of content to influence audiences is not new, but the line between what’s real and what’s fake has become razor-thin. Through disruption and deception, deepfakes can cause significant damage both to individuals and institutions.
GANs are capable of creating media that is virtually undetectable to cybersecurity experts. There are more than 100,000 models that have been developed to create and/or detect deepfakes, of which only about 3,000 can identify deepfakes properly. Most models are only able to identify the “Cheapfakes,” which are unsophisticated copies generated through the most simplistic manipulations.
Generative AI connection
Whether it’s Cheapfakes or Deepfakes, the economic model has shifted significantly because of Generative AI models. Recent technological advancements make it possible for anyone with a computer to create Deepfakes—and open-source code, free mobile applications, online tutorials, and inexpensive AI service providers have greatly expanded access to the necessary tools.
As a result, criminals are seeing a great leap forward in their ability to commit fraud, extort large sums of money, damage reputations, and even disrupt national security. AI is now being used to alter maps, imagery, and X-rays, generate text, and even create realistic artwork. And AI-generated digital avatars are increasingly able to hold conversations, exhibit emotion, and make realistic human gestures. The evolution of manipulated content has come a long way in a short period. Deepfakes are close to being readily available for a variety of purposes, not all of which are well intended.
If you'd like more information and to see where this information was retrieved from, please visit https://kpmg.com/us/en/articles/2024/deepfakes-real-threat.html