A massive dataset of deepfake videos was recently released by Google. But what are deepfakes?
Deepfakes refer to the videos that look highly convincing and almost authentic to untrained eyes but are tweaked to some amount.
These deepfake videos are generally used perceived as a nuisance are used to spread highly effective disinformation. Mostly these videos are known to target celebs.
Deepfake videos pose a significant threat in the upcoming 2020 elections.
But why would Google, a tech giant, would jeopardize itself to pull out some fun out of the internet? Well, here’s the full story.
Google filmed many actors in a wide variety of scenes and then used that footage to prepare some (actually many) deepfake videos.
Google used the tools publicly available while doing this. These videos were generated to aid researches focused on their detection and removal.
Researchers can now use this dataset and carry out researches to develop an automated detection tool which is as efficient and accurate as possible. This tools will focus on and target AI synthesized images.
Google is well acquainted with the fact that these kinds of technologies advance pretty quickly and has promised to add more videos to the dataset with the newest techniques in the market to keep the researcher updated.
Google said in an announcement “Since the field is moving quickly, we’ll add to this dataset as deepfake technology evolves, and we’ll continue to work with partners in this space. We firmly believe in supporting a thriving research community around mitigating potential harms from misuses of synthetic media, and today’s release of our deepfake dataset in the FaceForensics benchmark is an important step in that direction.”
People Also Read: PS5 will consume less power to Fight Climate Change
As these deepfakes are turning out to be a significant threat to the Cyberworld, Facebook has teamed up with Microsoft and some of the most prominent technological institutes including MIT and the University of Oxford in an effort to take down deepfakes.
Facebook plans to invest $10 million in a mission called DFDC standing for Deepfake Detection Challenge.
The need for a large Dataset was needed by a research conducted by the Information Sciences Institute (ISI) of the University of Southern California. In June researchers claim to have created an algorithm which, they say, can detect fake videos with up to 96 per cent of accuracy.
To develop their system more, they used more than 1000 manipulated videos before Google came out with its dataset. These tools are also used by bad actors to achieve more sophistication, so the researchers will need so more data.
Also Read: Manipulated videos that seem ‘perfectly real’ are coming soon!
Zao, an iOS-based face swapping app caused a missive turmoil which made the headlines recently. Zao quickly topped the Chinese edition of App Store since it was launched. But Zao had something in its user agreement that was noted by Bloomberg.
The agreement said that using the app gave the developer Changsha Shenduronghe Network Technology “free, irrevocable, permanent, transferable and relicense-able”.
In simple words, Photos and videos that contained your face now became the property of app creator.
While its good that these conglomerates are coming together to fight against cybercrime. Nothing can be said about the time of the success of this initiative
This website uses cookies.