Facebook Inc will be collaborating in a Partnership for AI coalition program with Microsoft Corp and academics from multiple universities to launch a competition program in its efforts to better coup up with detection of deepfakes, the company shared in its earlier this month blog post.
Aiming driving research around detection of deepfakes, the social media giant will conduct a “Deepfake Detection Challnge” by putting an amount $10 million into it. In order to test its detection tools, Facebook, as part of the project, is taking the researchers into service to produce realistic deepfakes to form a data base to apply its deepfake detection tools on to them.
The realistic deepfake videos, which will make no use of user’s data but will be prepared featuring paid actors, would be released in December, the company said.
With the U.S. presidential election ahead in November 2020, there has been building-up pressure on social media platforms to effectively handle the threats arising from the use of deepfakes, which make use of artificial intelligence in creation of hyper-relaistic videos in which persons came appearing doing or saying the things which they did not do or said in reality.
Though there has not been any well-created deepfake video surfaced so far that has largely impacted political scenarios in the United States, but recently a “cheapfake” clip of House Speaker Nancy Pelosi was recently used in demonstrating the potential threat arising out of such a manipulated video to cause confusion which was manually slowed down to make her speech seem slurred.
Facebook still lacks any specific policy against deepfake videos, but it’s Deepfake Detection Challenge is not for the first time when the social media giant came funding academic research to better tackle threats like these.
Earlier in July, Chairman of House Intelligence Committee, Adam Schiff urged Facebook, Alphabet Inc’s Google, owner of YouTube, and Twitter Inc to share their plans on how they are putting efforts to tackle deepfakes.