Microsoft has successfully developed a tool to detect deepfake images and videos online as it seeks to curb the spread of disinformation.
Synthetic media, also known as deepfakes, are images, videos, or audios that are manipulated by artificial intelligence to look real. These videos are often quite convincing which makes them hard to detect.
Microsoft’s new tool called Video Authenticator and it is able to tell with “a percentage chance or confidence score” that media on the internet has been altered artificially. It can provide this percentage in real-time as each frame of the video plays. It works by detecting the blending boundaries around the deepfake and subtle fading or greyscale elements that might be invisible to the human eye.
#Deepfakes no more. Behold, the Microsoft Video Authenticator, a tool that can analyze a still photo or video and provide a percentage chance that the media is artificially manipulated. (1/2) pic.twitter.com/IINud4lWmE
— Microsoft On the Issues (@MSFTIssues) September 1, 2020
The software was created with the help of a public dataset from Face Forensic++ and tested on the DeepFake Detection Challenge Dataset, both of which are leading deepfake detection and testing technologies.
Microsoft hopes that the new Video Authenticator will be used by organizations around the globe in the fight against disinformation. However, since deepfakes are powered by AI that can continue to learn and improve, Microsoft warns that it will eventually beat conventional detection technology.
Regardless, it should prove to be a helpful tool for detecting deepfakes for now.