The Ethical Evaluation of Deepfakes by Yizhen Wang

People used to believe that recorded videos were solid evidence, but with the rise of deepfake technology, this is no longer the case. Deepfake uses AI to swap faces in videos, making them look realistic. This technology has ethical concerns as it can be used to deceive and commit fraud. Deepfake apps have made the process easier and more accessible. Deepfake has been predominantly used for creating non-consensual pornography, damaging the reputation of celebrities and private individuals. It also poses a threat to businesses by generating misleading content that can harm their reputation and credibility. Another ethical concern is the disruption of social trust, as people may become skeptical of all information due to deepfake frauds. Governments will need to find innovative solutions to address these issues. People used to believe that recorded videos could be seen as a powerful evidence and that what they see in those videos must be true. However, this is no longer the case with the development of artificial intelligence and what's known as deepfake. The term deepfake actually comes from the underlying technology, deep learning, which is a form of artificial intelligence that is used to swap faces in photos and videos to make realistic-looking fake media. In today's podcast, I am going to be using specific examples of deepfake applications to show how such technology could actually impact our daily lives and why it raises ethical concerns. First of all, to demonstrate how powerful deepfake is, I have invited a friend of mine, Olivia, and I have already sent her a deepfake video to watch. But I didn't tell her that the video was actually altered by deepfake. Let's see how she would react. Hey Olivia, could you hear me? Hi Yizhen, yes, I can hear you. Alright, perfect. Thank you so much for joining me today. Now, could you please briefly describe your impression of the video? Sure. I think the video recorded a public speech given by the U.S. politician. I can't remember the specific argument that he made, but it was something about higher education, if I'm correct. Thank you, Olivia. That was great. Besides that, did you notice anything in the video that seemed a bit unusual to you? I'm actually not sure. I would say no because everything seemed normal. Okay, but what if I tell you that the politician's face was actually changed using AI technologies and who you saw in the video wasn't a U.S. politician, but just some random guy from the internet? Oh my god, that's totally unexpected. Honestly, I wouldn't be able to tell if you didn't tell me that. Wow, I still can't believe it. That is so magical, but it raises my concern because someone could play that trick to commit fraud. Brilliant. That is exactly what I'm going to be talking about later. Alright, thank you so much for your participation, Olivia. Now, although given the hint, Olivia did not notice the changed face at all. This demonstrates that deepfake can now generate very authentic-looking videos, but previously the transformation process needed to be done with the help of a powerful computer and that the whole process was actually quite complex. However, with the advancement of technologies, there are now plenty of mobile deepfake applications that largely reduce deepfake's barriers of entry for all users. For example, there are DeepFaceLap, FaceSwap, FaceApp, and so on, but an application called Zao was one of the first movers which went viral in China. Now, I'm going to invite another friend of mine, Lisa, who has previously used Zao application to share some of her thoughts on the deepfake apps. Alright, hey, Lisa. Can you tell us a little bit about your experience using Zao and how such technology makes you feel? Sure. I remember it was 2019 when I first heard of this application that can replace celebrity faces with my own. This sounds quite interesting, so I got the application right away to have a try. All I needed to do was to insert pictures of my own, and all of a sudden, I become the actress in the scenes. For me, it was so much fun. Great. Now, the other question I have for you, Lisa, is while using this for fun, were you somehow concerned that the application usage will potentially exploit people's right of image and infringe their copyrights? To be honest, I did not pay much attention to the agreements I had to sign before using the application, but I feel scared when I understand how deepfakes can facilitate the creation of misleading contents and pornography that damages women's reputation on purpose. Alright, thank you, Lisa. You're absolutely right. Although deepfake has certainly brought so much fun across social media platforms, it has great potential for sure. It is obviously a dangerous technology with some troubling applications, including the synthetic pornography that you have just mentioned. Pornography actually made up the largest proportion of all content generated by deepfake, and that is over 90% of all deepfake videos found online since 2018, according to a research company called Sensity AI. This demonstrates how severely deepfake damages the reputation of celebrities and prominent figures. Well, there's another deepfake pornography application called Deepnude. It is basically a photo-generating software that emerged in the US in 2019, and it allows users to upload a female image and then uses deepfake algorithms to automatically generate a new version of the original photo uploaded. The whole transformation process would only take a short amount of time, and it does not require users to have any technical skills at all, making it so easy for them to create pornographic content with just a few clicks on any individuals they want. And this time, the targets were not just celebrities and online influencers, but also private individuals and even underage girls, where at least 200,000 victims were reported by 2021. Overall, the use of deepfake allowed the creation of non-consensual porn targeting women and also enabled sexual violence against women. However, besides the ethical concern around pornography, deepfake also raises other ethical concerns as well. It potentially damages the intangible assets of enterprises, including their brand goodwill and brand reputations. As we know, for businesses, it is definitely crucial to have a great credibility to be able to thrive in a market economy. Therefore, all enterprises seek to improve the external evaluation of their products, operations, and services by society. But deepfake has become a tool for individuals and organizations to discredit companies by generating misleading content, because it's increasingly difficult to tell whether videos and audios are authentic or they're just generated by deepfakes. For example, when they create videos supporting that an executive of a company commits legal or ethical violations, such as making racist expressions, having inappropriate relationships with others, and bribing government officials, these misleading contents can be published to target a company at its key commercial stages, such as merger and acquisition negotiations, initial public offerings, or bidding participations. Although these videos would eventually be proven false, they would still damage the company's business reputation, eliminate potential business opportunities, and even result in a serious loss of assets. An example would be when a deepfake video was released suggesting that the CEO of a Chinese state-owned company was involved in a public scandal. This has caused a significant decrease in the company's share price, although the content was later proven to be false. We can now look at deepfake's ethical issues on infringing individual image rights through synthetic pornography and damaging companies' goodwill through intentional accusation. However, the third ethical concern of deepfake is even more significant, which is the potential disruption of social trust, where people would find it difficult to believe in anything and view all information as suspicious because of multiple types of frauds. Well, here's another example. A UK-based energy firm encountered a deepfake fraud in 2019. According to the Wall Street Journal, the criminals used the deepfake technology to imitate the voice of the company's senior manager, who successfully required the company to send 220,000 euros to a supplier. Imagine what would happen in the future when the technology becomes even more sophisticated. Deepfakes would result in people being skeptical of all information. For example, truths may not trust official comments, and the public may believe that the serious scandals and outrages are fully all fake news. Politicians and their supporters will be able to reject anything negative about them as false or exaggerated. From the three ethical issues of deepfake that we have discussed so far, it can be seen that the governments would have to come up with innovative solutions that address the potential harms of deepfake. Well, because of its complexity, the solution in the future will have to be an integration of technical, regulatory, and social approaches. That is everything in today's podcast. I need to say thank you very much to the two guests, Olivia and Lisa, who have given their brilliant insights. And thank you all for listening.