Nowadays, it’s easy to put words in someone’s mouth. How? Just with deepfakes, which are the specific kind of synthetic media. Tools supported by artificial intelligence that allow creating compelling images, audio, and video hoaxes have become so familiar and accessible that they are a severe threat to celebrities or politicians and all of us.   

The concept of fake news is already well known to us – it is false information shaped in such a way as to arouse emotions, be digestible, and be easy to pass on. Deepfake, on the other hand, introduces an informational lie to a higher level – thanks to artificial intelligence (AI), images and recordings are created that look as realistic as possible but have nothing to do with reality. Why is it called deepfake? It’s a made-up word coined from the words “deep learning” and “fake.” To make a deepfake, a video maker changes one person’s face and replaces it with another, using a facial recognition algorithm and a deep learning computer network. Who can create the deepfakes? From amateurs to academic and industrial professionals, visual effects studios, and movie producers.

Have you seen Queen Elizabeth II dance on the table while delivering a Christmas speech? Or have you listened to Kim Jong Un’s lecture on democracy? Deepfake technology is often used to create convincing yet entirely fake video content and can be exploited to manipulate facts. Learning algorithms can develop messages that are very difficult to distinguish from reality. They are easy to share, and they can spread panic. That’s what their creators might want. 

On the other hand, deepfake technology brings possibilities to multiple industries, from marketing to museum and education. Deepfakes can save time, and time is money. In the movie industry, deepfakes can help with editing video without a reshoot. They can also help to recreate the past and bring history and art to a broader audience. A great example is Dalí Lives, the initiative of the Dalí Museum in St. Petersburg, Florida. Using the machine learning-powered video editing technique, the museum visitors could learn more about Salvador Dalí by interacting with Dalí himself. 

Fears over the malicious use of deepfake technology bring us the question of how to detect them. It gets more challenging as the technology improves, but some deepfakes with poorer quality can be pretty easy to spot for us. Pay attention to the face, audio and lighting. Does the person blink enough or too much? Do their eyebrows fit their face? Is someone’s hair in the wrong spot? Maybe the skin is too patchy? The lip-syncing might be wrong, and the voice does not match the person’s appearance. Do eyes have the fitting reflection? You do not need to become a deepfake expert to distinguish what is real from what is fake, but you should remember one rule. Think before you share. The spreading of counterfeit videos may have irreversible consequences.

Today’s wars do not just take place in reality. No less critical battle is fought on the Internet: for truth and reliable information. Deepfake is just one of the examples of the synthetic media phenomenon, which is driven by ever-improving AI and the growing possibilities of 3D graphics. We need to be more vigilant about what we trust on the Internet, whether we want it or not: technology is a double-edged sword.

Jolanta Ciopcińska

Sources:
YouTube Channel 4: Deepfake Queen: 2020 Alternative Christmas Message
YouTube Represent Us: Dictators – Kim Jong-Un
MIT Sloan: Deepfakes, explained
The Dali: Dalí Lives