You know that deepfakes can be devastating to one's reputation and are very difficult to detect if created properly. This problem was thought to be faced by Microsoft releasing a tool that can detect them.
The threat of deepfakes is so significant that it did Facebook to ban any deepfake content. However, the biggest obstacle is how if the creator has knowledge they look very convincing. So the localization and confirming a deepfake is very difficult.
Facebook bans deepfakes and Photoshopped photos because they deceive users.
As mentioned in Microsoft On the Issues, the company wants to change that with a new tool. It's called Microsoft Video Authenticator and it analyzes video to look for the "seams" left behind in the creation of a deepfake. These seams are invisible to the human eye, but one computer can detect them using frame-by-frame analysis.
To prevent any reverse engineeringς της νέας τεχνολογίας, το Microsoft Video Authenticator δεν κυκλοφορεί δημόσια. Η Microsoft διανέμει την εφαρμογή της μόνο σε οργανισμούς ειδήσεων και πολιτικά κόμματα. Έτσι μπορούν να χρησιμοποιήσουν το εργαλείο για να διαχωρίσουν το πραγματικό από το ψεύτικο.
However, Microsoft has announced a second technology for detecting deepfakes. The company proposes a system where a content creator can add tags to his videos with a unique "signature". If the video is edited, a different signature will be created than the original, specifying deepfake.
Deep Fakes and the Infocalypse author Nina Schick states:
The only really common use we have seen so far is non-consensual pornography against women.
However, they are expected to be ubiquitous in about three to five years, so we need to develop these tools for the future.
However, as detection capabilities improve, so will the ability to create.
Deepfakes are a major issue for media credibility and companies need a way to separate them.
Microsoft has released the tool to detect edited videos, however, we do not know if it is enough to locate deepfake creators it has not yet seen.
