What are Deepfakes?
Deepfakes, a combination of “deep learning” and “fake”, are hyper-realistic videos, images, and sounds that are digitally manipulated to create people or events that do not exist, or to depict people saying and doing things that never actually happened. Deepfakes rely on neural networks that analyze large sets of data samples to learn to mimic a person’s facial expressions, mannerisms, voice, and inflections. The deepfake video process involves feeding footage of two people into a deep learning algorithm, using facial mapping technology and AI, in order to train it to swap the face of one person on a video with the face of a different person. Deepfake videos are difficult to detect, as they use real footage, can have authentic-sounding audio, and they are often optimized to spread rapidly on social media. Deepfakes can be created using the following1:
- Autoencoder. An autoencoder is a special type of neural network that is trained to copy its input to its output. For example, given an image of a handwritten digit, an autoencoder first encodes the image into a lower dimensional latent representation, then decodes the latent representation back to an image.
- Generative Adversarial Networks (GANs). Two models are trained simultaneously by an adversarial process. A generator learns to create images that look real, while a discriminator learns to tell real images apart from fakes. During training, the generator progressively becomes better at creating images that look real, while the discriminator becomes better at telling them apart. The process reaches equilibrium when the discriminator can no longer distinguish real images from fakes.
Deepfakes are often used for the following purposes2:
- Scams and hoaxes. Cyberattackers can use deepfake technology to make false claims, create scams, and to construct hoaxes designed to undermine and destabilize organizations.
- Social engineering. Deepfake technology can be used in social engineering scams. Audio deepfakes can trick people into believing that trusted figures have said something that they did not say in order to prompt them to take a desired action, such as transferring money or divulging sensitive information.
- Identity theft and financial fraud. Deepfake technology can be used to create new identities and steal the identities of real people. Cyberattackers can use deepfake technology to create false documents or fake their victim’s voice, which enables them to create accounts or purchase products by pretending to be that person.
- Election manipulation. Deepfake videos have been used to spread fake videos of world leaders, in order to influence elections by manipulating voters.
- Nonconsensual pornography. Nonconsensual pornography accounts for most of the deepfake videos on the internet. These often target celebrities, but deepfake technology is also used to create hoax instances of revenge porn.
- Automated disinformation attacks. Deepfakes can also be used to spread automated disinformation attacks, such as conspiracy theories and incorrect theories about political and social issues.
1 OWASP, 2022, “Deepfakes: A Growing Cybersecurity Concern”
2 Fortinet, 2023, “What Is a Deepfake?”