In the autumn of 2017, a Reddit user going by the name “Deepfakes” posted several pornographic videos to the Internet. Well-known actresses like Daisy Ridley, Gal Gadot, Emma Watson, Scarlet Johansson appeared in these videos. However, all of these videos turned out to be fake. They were generated by Artificial Intelligence, hence the name “Deepfake”; a blend of “deep learning” and “fake”. It did not take viewers long to realize these were not real videos, and they were taken down shortly after. Ever since then, this technology has been improving. Some websites, such as Twitter and Gfycat, are actively combatting these types of videos by removing them from their platforms and proactively blocking the uploaders.
How does it work?
The Machine Learning algorithm analyses millions of frames of a particular face of the well-known actresses mentioned before. After the algorithm has learned the key features of this face, a source video or image is fed to it. The algorithm will then overlay the analyzed face onto the source video and mimic the source’s movements, resulting in a very realistic Deepfake video.
Dangers
Although picture and video editing has been around for a very long time, the new danger with Deepfakes is in the realism of these videos. The owners of these systems can use them to create shocking videos of influential people, with the public unable to distinguish them from reality.
We are all familiar with Fake News, and deepfakes are creating an extra dimension to this dangerous phenomenon. We could be heading towards an era where nobody knows if anything in the media is true anymore. People could also use this in their favor, claiming that a particular comment or action was not them, but actually a Deepfake video.
Actor Jordan Peele collaborated with Internet media and news outlet Buzzfeed to issue a public announcement on the dangers of Fake News. They did this by creating a Deepfake video of Barack Obama. Watch the video below; at 00:35 the source video is shown next to the result.
What do you think, how do we combat an issue like this?
Sources:
https://en.wikipedia.org/wiki/Deepfake
https://www.theverge.com/tldr/2018/4/17/17247334/ai-fake-news-video-barack-obama-jordan-peele-buzzfeed
Hi Daan,
Thank you for the interesting read you provided. I am personally quite worried about the possibility of creating deep fakes. At this stage, usually celebrities are targeted, and creating a ‘good-looking’ deep fake is still only possible if you know how to get the right tools and how to use them. However, as technology is emerging quickly, I wonder how long it will take before we can download extremely well-performing Deep Fake Apps in our Play Store or App Store. This could easily lead to big troubles and misunderstanding. Maybe you have some ideas on the measures Google/Apple could take to prevent this from happening?
Hi Daan,
Thank you for the interesting blog! I was attracted to your blog by the tittle. It was the first time that I heard about the notion deepfake. Artificial intelligence is now becoming better and better. It is indeed difficult to distinguish between what fake is and what not. Seeing the video made me wonder I how long it will take that we cannot distinguish what is real and what not. The technology is growing fast and changing our lives. So, I see the danger here. What are your thoughts about it? Do you think that there is a way to prevent this?