Vishal Asnani, PhD researcher at Michigan State University, teamed up with Meta to develop a method for proactive detection of Image manipulation, or deepfakes. Deepfakes refers to computer-manipulated images or videos which appear to be authentic, and are often leveraged with malicious intent on social media platforms to defame social figures, and re-imagine civic events from a fictitious perspective. These altered artifacts are difficult to detect and in prolific circulation across the internet. In recent history, deepfakes have been charged with disrupting political affairs and elections and are thus top of mind for many, but especially one of the world’s most popular social media outlets, Meta. Asnani will present his research which was presented at CVPR 2022.

GitHub: Reverse Engineering Generative Models