Image Rights byNMashinini

Deepfakes and Image Rights

April 30, 2021 Nomalanga Mashinini Season 1 Episode 4
Image Rights byNMashinini
Deepfakes and Image Rights
Show Notes Transcript Chapter Markers

In this episode I discuss how deepfakes impact image rights.

Updates on Image Rights are published on social media @irbynmashinini


Host: Nomalanga Mashinini

Cele v Avusa Media Ltd [2013] 2 All SA 412.

Chesney & Citron, ‘21st century-style truth decay: Deepfakes and the challenge for privacy, free expression, and national security’, (2019) 78 Maryland Law Review 882.

Farish, ‘Do deepfakes pose a golden opportunity? Considering whether English law should adopt California’s publicity right in the age of the deepfake’ (2020) 15(1) JIPLP 40.

Ferraro, ‘Deepfake legislation: A nationwide survey – state and federal lawmakers consider legislation to regulate manipulated media’ (2019) WilmerHale, available at accessed on 26 July 2020.

Daily Maverick article on deepfakes by S Grootes Accessed on 17 April 2021. 

Jordan Peele example of deepfakes using Former President of the United States of America, B Obama accessed 24 April 2021.

Le Roux v Dey (CCT 45/10) [2011] ZACC 4.

Nick Cage Deepfakes accessed on 24 April 2021.

 We are a growing community seeking to ensure the protection of Image Rights for all people. Thank you for contributing to the growth of this podcast.

Support the show

Welcome to another episode of Image Rights by N Mashinini. I'm your host, Nomalanga Mashinini, and I will be sharing my knowledge and research on the legal and contemporary issues that relate to image rights.

I recently saw an article on deepfakes in the Daily Maverick with the title that says, "stop believing your lying eyes". This article speaks about the arrival of deepfakes in the South African political arena. You see, deepfakes have rapidly entered our lives. But some people are still asking what are deepfakes? So in this episode, I explain deepfakes and how they violate our image rights. When I saw that article in the Daily Maverick, I paused to wonder, is it really our eyes deceiving us? Or is it the technology that is used to create deepfakes that is deceptive?

You see, deepfakes are audio visual media, such as photographs, videos and audio, just that, they're fake. Although they look and sound convincingly real, deepfakes are created using deep learning algorithms, which are basically a subtype of artificial intelligence machine learning. When you hear that deep fakes are created by AI, you may assume that this technology is not freely available to the average 'Joe'. But this is exactly the problem with deepfakes. The software used to create deep fakes is often free, and has really good quality. In fact, a deepfakes mobile app called REface is available in South Africa as we speak. It has better quality than the old Face Swap, just that it's creepier.

The problem with deepfakes is that these images or videos and audio recordings contain people's identity features like their faces, their body parts and voices. And these parts or identity features are used to create media files of something that may have either never happened before, or to recreate an existing audio or clip of someone. If you look on YouTube, for instance, you'll see that deepfakes have become quite popular they, as you'll find multiple deepfake videos of Nick Cage doing various stunts that he never actually did, and also one of Barack Obama super-imposed on Jordan Peel's face. These are videos of ordinary people, and sometimes even comedians that are replaced with the images of these public figures to make it seem as if they had either said or did something that they actually didn't. And sometimes deepfakes, use clips from movies, and replace the images of the stars with that of their fans. Even worse, deepfakes are commonly used for revenge pornography. And they're quite used very often for purposes of touching people's reputation in the public.

You may have heard of the incident with Congresswoman Stefanik last year, where there was a deepfake circulating on social media, showing her putting up a middle finger while she's on duty, which actually never happened. It took a couple of hours for people to even notice that this was a deepfake. Because you see, it takes a good eye to pick up that video or photograph, even, is fake. The problem is, however, that social media users are not attentive to what is real or not. Once a deepfake spreads, it is hard to remove online.

When people listen to a recording thinking that it's real, they often believe it without asking further questions, or when they watch a video or see a photograph they quickly share it with their friends on social media, and it goes viral. Professional media reporters are probably the ones who are more cautious to investigate the correctness or accuracy or reality of certain material. But as for the rest of the social media users, many are not aware of the deceptiveness of technological innovations such as deep learning algorithms. The average social media user is more likely to believe what they see and hear, and this kind of online behavior threatens our image rights.

Deepfakes are insulting in nature, especially when they're targeted at ruining someone's reputation, especially when their used for obscene purposes without the permission of those who are victimized through deepfakes. As a result, often times women are victims of deepfakes, especially with pornography cases. Deepfakes are often also used to objectify people and embarrass them in public. They injure feelings and degrade others. And this means that deepfakes can also sometimes be defamatory. Authors like Kelsey Farish, Matthew Ferraro, Chesney and Citron explain at great length, all the legal ramifications that come with deepfakes but they strictly speak to issues relating to the United States of America.

South African law does not outright say anything about deepfakes. Although we do have a system of criminal and civil law, in this podcast episode, I'm discussing the civil law aspects that relate to deepfakes and image rights. You see, in terms of the civil law issues, it's mainly about personal or private relationships between persons, it's not really related to crimes or anything like that. So when one sues another person for something like defamation, it's dealt with through civil law. In this way, you and I could sue anyone who uses our images, or even voices in a deepfake, without our permission on the basis of personality rights, such as the right to privacy, the right to good name, the right to feelings, the right to human dignity, and, of course, image rights. I believe that deepfakes affect all of these rights in some other way, depending on the circumstances. But in my view, the most prominent rights that are affected by deepfakes are image rights.

Because remember, image rights directly protect your ability to control who uses the aspects of your identity, and how they do so in the public domain. But deepfakes go right against that bundle of image rights in that they exploit your photographs, your voice, and sometimes even other parts of your body, depending on the context. So when a person inserts your voice or image in a deepfake audio or video recording, they are directly using your identity for their own purposes and for their own benefit against your will. So this way, your image rights are violated by the misrepresentation that is created to the public that it makes it look as if you've maybe consented to your image being used in a deepfake. It may even look as if you participated in certain questionable activity that is depicted in the deepfake when you actually didn't. Either way, the fact that the deepfake casts you in false in a false light calls for legal intervention because the deepfake is depicting circumstances that never actually happened.

And so when deepfakes are also used on mediums that can generate income for user content, then the issue becomes one of commercial exploitation. Because, then a person who has created and distributed a deep fake of you is actually using either your face or your voice or your body parts, or any identification feature to make money for themselves. In South Africa, you can sue a person for that kind of activity when they use your identity for the commercial advantage. It may be harder though, if you're a public figure, especially when deepfakes are used in instances where it constitutes parody. This is just in line with the right to freedom of expression.

Deepfakes also remind me of a very famous case that happened in 2006, although it was heard and decided much later than that. And that is the case of Le Roux v Dey, where this judgment actually went all the way to the Constitutional Court, and a high school pupil had placed the faces of the deputy principal and principal on another image of two bodybuilders who are posing in a sexually suggestive manner. And in that judgment, the majority was not sympathetic to the fact that the child was just creating this image for purposes of amusing themselves as well as their peers. In fact, Dr. Day sued for quite a hefty amount of money, although he was hoping for about R600,000, if I remember correctly, he only received 25,000 In compensation, but there was a lot more punishment for the school children who were involved in this case, because there was also a criminal matter that was brought against them. And they eventually had to go and do some community service at a local zoo. They were also detained at school, and deprived of participating in certain prestigous awards at school, and they were forced to give an apology to the deputy principal. So it is quite clear what the consequences of creating and distributing a deepfake are in South African law, because the principles that were applied in the rule versus day case will obviously apply even today in the context of a deepfake.

Although it becomes a bit difficult when you're dealing with issues of how many people are liable, because of social media, when a deepfake goes viral, there could be 1000s of people who are involved in distributing, sharing, liking and posting that particular deepfake. So who do you sue? How many of them do you sue? Who is actually liable? It becomes a bit difficult to determine who is actually liable, it also becomes very difficult to find out who created the deepfake.

The judgment of Isparta v Richter, is a very important one for this particular position because in that judgment, the court actually held that in defamation cases, and obviously also in other personality rights issues. anyone who is tagged in a defmatory post on social media will be jointly and severally liable with whoever posted it initially. That is interesting, because if I do not remove my name from a deepfake posts when I'm tagged to it, simply because I'm not aware that it is a deepfake, therefore, meaning that I'm also deceived that this is a real post, it becomes a bit unfair to hold me liable for that particular deepfake along with whoever created or posted it first. I'm seeing that deepfakes are going to challenge our legal principles under the common law a little bit. But the foundational issue that remains is that deepfakes are punishable under our laws, and they're soon going to be even more punishable in the criminal context with the incoming of the Cyber Crimes Bill.

So if you want to learn more about deep fakes and the ways in which South African law is changing to combat deep fakes, and other cyber crimes, you can tune into the next episode, I will be speaking about deep fakes and the Cyber Crimes Bill.

In the meantime, subscribe to Image Rights by N Mashinini. I'm your host, Nomalanga Mashinini, a lecturer at Rhodes University.

To keep up with more episodes, you could also follow @irbynmashinini. That is the official podcast page.

And remember, this podcast series is based on research supported by the National Research Foundation of South Africa, grant number 121887, but the views and opinions expressed in this series do not reflect the views and opinions of National Research Foundation, its management or its governance structures.

Until next time, control your image rights!

Keywords: deepfakes, identity, cybercrimes, social media, image rights, publicity. 

Deepfakes arrive in SA elections
Definition of Deepfakes
The problem with deepfakes
The Stefanik middle finger deepfake
South African law regarding deepfakes
Image Rights violations through deepfakes
Deepfakes for gain
Le Roux v Dey
Isparta v Richter