A deepfake is a type of artificial intelligence technology that uses deep learning to alter images or videos by putting someone else’s face onto the face in the original content. Doing this to media creates a puppet-like situation, making someone appear to say or do actions they are not doing in the original source. The term ‘deepfake’ is derived by combining “deep learning” and “fake” and the technology is used to produce or alter video and audio content so that it shows something that didn’t occur.

Before deepfakes appeared in 2017, special effects used in movies like Jurassic Park and The Matrix enabled audiences to suspend reality and enjoy fantasy on the big screen. Movie audiences willingly agreed to suspend their belief in reality when they purchased a ticket so the line blurring reality with fantasy was not fooling anyone. Movie special effects require a skilled production team and the use of expensive software to create its final products.

Deepfake technology has changed this paradigm because special skills and expensive software are no longer needed. You may have seen deepfakes in news stories about politicians, tech executives and celebrities, and many of them look very realistic. So far, the most widely seen deepfakes have been satirical and comedic, however, they have the potential to be nefarious. For example, a deepfake could falsely show a politician doing something reprehensible in order to sway an election or a schoolyard bully could harass a classmate who’s been targeted.

Lawmakers and regulators are just beginning to grapple with the legality of deepfakes. It is not illegal to produce a deepfake but Congress is considering making it unlawful to maliciously create and distribute them. There is urgency to have legislation enacted prior to the 2020 election because there is no foolproof technology that can reliably and consistently identify videos as being deepfakes.

How easy is it to produce a deepfake without using special equipment or creating a special environment? I decided to give it a try using two video clips of comedian Will Farrell impersonating President George W. Bush and a clip of President Bush speaking. Using a free face-swapping deepfake site, I spent 10 minutes configuring the settings before letting an algorithm on the site track each face so that it could map one face onto the other. The next morning, my deepfake video was ready.  Because I didn’t want to spend much time and any money, the results are a little less realistic than some of the other deepfakes that I’ve seen online. However, if someone thought that they were watching an actual video of President Bush, this deepfake would make the video seem more realistic. Check out my deepfake video on YouTube, An example of how deepfakes enhance the realism of a video.

Curated below are several great resources that discuss deepfakes, the main types of manipulated videos designed to fool, the harm they create, ways to tell if a video may be a deepfake or not, and what you can do if you become the victim of a deepfake.

• First, let’s take a closer look at what deepfakes are, how they’re becoming more realistic, and their potential to mislead as well as entertain, by watching a short video produced by NOVA PBS entitled Deepfake Videos Are Getting Terrifyingly Real.
• Become a more aware consumer of content by watching The Fact Checker – How to spot manipulated videos from the Washington Post. This video identifies, describes and provides three examples of the main types of manipulated videos that are designed to fool people.
• Now that we know more about what deepfakes are, how can you tell if a video that you are watching is a deepfake? Kyle McDonald offers nine tips about how to tell if a video could be a deepfake in his blog How to recognize fake AI -generated images.  Today, it is not easy to tell real from fake and as deepfake technology continues to improve, it will become even more difficult to do in the future.
• Finally, Bernhard Warner’s Fortune article Artificial Intelligence Is About to Make Ransomware Hack Attacks Even Scarier and its accompanying video does an excellent job explaining deepfakes, the tension between free expression and law and decency that protects people, and offers suggestions about what you should do if you become the victim of a deepfake.

If you have questions about deepfakes, email them to danielkent(at)netliteracy.org and visit NetLiteracy.org, a tech oriented digital literacy and inclusion nonprofit that developed a new program called AI Literacy that benefits and provides resources for teachers, parents and teens.