Facebook AI Software can find the Origin of Deepfake Images

Deepfakes term is used for photos, videos, or audio clips that are manipulated and made to look authentic by using artificial intelligence practices and techniques. Experts have warned already about such content. It is manipulative and can mislead to wrong perceptions. Scientists at the Facebook research center developed AI-based software that will use reverse engineering to find the authenticity of the images, videos, and audios. It can also backtrack the origin of the content.

In a Blog post Tal Hassner and Xi Yin, a scientist at the Facebook research center, told the audience that their team, teamed up with Michigan State University to create this software.

“Our method will facilitate deepfake detection and tracing in real-world settings, where the deepfake image itself is often the only information detectors have to work with,”

They further added “This work will give researchers and practitioners tools to better investigate incidents of coordinated disinformation using deepfakes, as well as open up new directions for future research,”

Pic Credit: Facebook

This new software by Facebook runs deepfakes on a network to find the imperfections left while manufacturing process, which also alters the digital “fingerprint” of the images.

“In digital photography, fingerprints are used to identify the digital camera used to produce an image,” the scientists said.

“Similar to devise fingerprints, image fingerprints are unique patterns left on images… that can equally be used to identify the generative model that the image came from.”

“Our research pushes the boundaries of understanding in deepfake detection,” they said.

Similar software was released by Microsoft last year ahead of the US presidential elections. The purpose was to counter hard to detect deepfake photos and videos.

This video Authenticator software was based on analyzing images or frames of the videos to find evidence of manipulation. An evidence a naked human eye can not detact.