Deepfake is a new and emerging tech that consists of AI-generated videos of a person edited with another person’s voice/face. Deepfake applications came out in October 2019, but the number of users rapidly increased in the recent pandemic up to 300 percent.
Deepfake tech can prove to be very harmful since anyone can have access to it and it can be used to tarnish anyone’s reputation may be due to political or personal reasons. This tech is already being used to make fake pornographic material that too without the consent of another person.One thing that is a matter of serious concern is that with such advanced technology at everyone’s hands it can be pretty difficult for anyone to differentiate between the real and the fake videos.
According to the researchers at Sungkyunkwan University in Suwon, South Korea, even the application programming interface (API) from Amazon and Microsoft can be deceived with the Deepfake generated videos. It was also said that the API at Microsoft’s Azure cognitive services had been double-crossed by the Deepfake generated content by almost 78 percent.
Till now it is quite difficult for Microsoft and Amazon’s APIs to distinguish the deepfake impersonator from the genuine content. This can be handled by creating suitable defense mechanisms and better designing web-based application programming interfaces.
Recently the Hollywood actor, tom cruise of the mission impossible franchise, got his Deepfake video made on TikTok which went viral.
The researchers are continuously working on finding the weak points of the deepfake detection AI. Up till now, the commercial facial recognition API has been tested on 5 data sets, two are made by the researchers and 3 that are made publicly, and it is found that the public figures whose pictures are videos are in bulk on the social media are more prone to get created as well as in an easy way.
Read next: These Are the Countries Producing the Most Electronic Waste
