Uses
Various types of deepfake technology and how they are implemented
Face & Voice Swaps
Face swaps use Deep Neural Network (DNN) technology to swap faces onto bodies using preprocessed samples of the donor and target.
​
Trained networks are then fed the target's video/image frame by frame, which is then decoded by the donor's decoder network
​
Many social media platforms utilize face-swapping technologies, though each platform varies in its use of technology.
​
Face swaps are one of the most common techniques to create deepfakes.
Department of Homeland Security, Increasing Threat of DeepFake Identities
Videos & Images
Artificial intelligence softwares are fed with information, then reworked until the generators are able to produced perfect replications of the content.
​
Videos and images produced with deepfake technology can pose serious threats, as it violates the notion of personal security.
via HS Insider
Generative Text
via ADL
Generative Artificial Intelligence includes software like ChatGPT, Siri, Alexa, and Speechify.
​
These AI sources are able to generate summaries of existing content, create replications of images or videos, and produce new content from target sources.
These models can also be used to create social media bots to promote propaganda or other spam content
​
via Magna AI
via Robot Named Roy
The following videos show the development of deepfake videos within a year. The top video was produced in 2024 and the bottom video in 2023.
​
In less than a year, the quality of deepfake videos has grown exponentially.
​
Most produced content is still somewhat differentiable from real videos. It seems as though AI struggles with perfecting fluidity in motion, though it continues to improve.
​
Yet, with the increasing accuracy of deepfake and AI-generated videos, it is important to take into consideration the risks that arise, as well as how society has attempted to mitigate those risks.