The Department of Homeland Security recently wrote a report highlighting the fact that Deepfakes pose a major threat to National Security, quite possibly the biggest threat we face today.
Deepfakes are computer generated videos that can take footage of a person and make them appear to say things they have never said. These videos were easy to identify as fake when they first came out, but now they’re becoming indistinguishable from the real person.
Subscribe to PBD’s YouTube channel today for every episode LIVE or On Demand!
The concept of a Deepfake is nothing new. Misinformation via video clips existed as early as 1898. The Edison Company wanted to film the Spanish-American War, but the low quality cameras made it challenging. Edison chose to mix real footage of marching soldiers and weaponry with staged footage of American soldiers.
The first version of modern Deepfake technology was the Video Rewrite Program (1997).
- This is one of the earliest programs used for modifying existing video footage. The program would first track facial movements in the video.
- Once the facial model was established, the program could then ‘rewrite’ the video with the synthesized facial expressions.
- While this was groundbreaking at the time, the synthesized video was not as seamless or realistic as what we see with more modern techniques.
- The animation is mainly 2D and doesn’t account for a lot of the complexities of human face and expressions.
The Deepfakes used today are called Generative Adversarial Networks (GANs).
- GAN consists of two parts: a generator network, which creates the images, and a discriminator network, which tries to tell the difference between real and generated images.
- The two networks are trained together, with the generator trying to make images that the discriminator can’t tell are fake, and the discriminator getting better and better at telling the difference.
- Over time, this results in the generator producing highly realistic images. Deepfakes created with GANs can be very convincing, as they can generate new images in 3D and handle variations in lighting, angle, and facial expression.
Famous examples of recent Deepfakes have been Barack Obama, David Beckham, Nancy Pelosi, Tom Cruise, and Joe Rogan. Earlier this year there was a Deepfake of President Joe Biden saying that there was going to be a draft in the U.S. to help Ukraine fight Russia.
As Deepfakes get better and better, they will pose an increasingly serious threat to National Security. This can range from political misinformation to Blackmail. People need to be on high alert and constantly be mindful of this technology and the fact that what they’re watching might not be real.
Add comment