November 10, 2024

Opinion: The Ethical Challenges of AI Deepfakes

By Samantha Amaro
@redlineproject

Art direction by Samantha Amaro | Illustration by MidJourney

AI Disclosure: MidJourney was used to generate the images in the article. And, the summary and podcast were generated by Summarize Wise Custom GPT in ChatGPT, and NotebookLM was used to generate the podcast. Lastly, the video was created using AI.INVIDEO.IO.

Summary: The paper explores the ethical concerns surrounding AI-driven deepfakes, particularly in generating non-consensual content that exploits and harms individuals, especially women and minors. It highlights cases from South Korea where deepfake technology is used to create and share explicit content anonymously, often with little legal accountability. The paper argues that current laws inadequately address the harm caused by deepfakes and calls for stricter regulation and public awareness. It emphasizes the role of platforms in managing AI misuse and advocates for ethical guidelines and stronger protections to prevent deepfake exploitation and support victims.

The creation of Artificial Intelligence (or AI) is not something new. We have seen the rampant use of AI by people with dozens of digital tools such as Google, Meta, Twitter, and others for more than a decade. According to the following “as of early 2023, some emerging generative AI systems had reached more than 100 million users.” We can see the reach that AI has on the general population, and the different uses the technology provides from content creation to “text, video, audio, codes and images.” However, when one thinks of AI they may perhaps think of software like ChatGPT, or chatbots such as Gemini, and not the fact that it can be used to produce deepfakes.

Deepfakes are synthetic media that can take a person’s image to create manipulative content, and their prominence is increasing. Status does not matter: normal people, celebrities, and politicians are all capable of being victims of deepfakes. This technology allows people to create scenarios that never happened, they can alter people’s image and misconstruing their words, and it is harder to tell what is real and what is fake. Although you can use deepfake technology to create your favorite artist covering a song, to create concepts for their work, others can create videos of your favorite artist shoving fans at the airport. There are loopholes in AI technology that allow everyday people to create deepfakes and face no repercussions, and this technology is being used to sexually exploit women. Throughout this essay, I will discuss the harm and question how much ownership one has over their self, i.e., face, voice, and body. Then I will talk about how deep-fake technology is being used to create explicit images, videos, and speech, and how there are more regulations over the extensive use of deepfakes.

How are Deepfakes Being Used?

Deepfake technology is being used to sexually harass women, particularly in South Korea. According to a report in The Guardian: “The National police agency said this week that it was investigating 513 cases of deepfake pornography — in which the faces of real women and girls are digitally superimposed onto a body without their knowledge or consent.” Women throughout the country are being exploited for existing. This type of pornographic content is being uploaded onto websites with restricted access and is being spread amongst people in private chat rooms. One main perpetrator of this spread is the app Telegram: Telegram, an app known for its anonymity and inability to track down users, has become the center of its distribution of child pornography, and “over its role in enabling fake pornographic images of children to be distributed. We can see how anonymity is being used to create and spread deepfakes, which allows its users to face no repercussions for their actions. More importantly, Telegram offers a self-destruction mode that allows users to leave no trace of messages they have sent and links that are being shared between users. This can leave victims of sexual assault on edge because, at the point of destruction, they cannot know the extent to which their image has been spread.

PODCAST: Listen to an AI-generated podcast of this article.

Dismissal of Deepfakes

Furthermore, “Deepfake videos may be dismissed as mere pranks, but they are criminal acts that exploit technology under the shield of anonymity, responsibility to report and investigate fake-nude cases largely falls to schools.” Aside from adults who engage in this type of exchange, teenagers are also the ones creating deepfake videos, and because it is termed as a “prank” it leaves room to question if they know the impact of what they are doing. There are currently no laws against “buying, storing, or watching deepfakes,” and “you find a chat room on a holiday, but by the time you get court approval, it’s gone.”This shows the permissibility of deepfake and allows for no accountability of perpetrators for the harm that they are causing. And, that there is knowledge they are being watched, which allows for them to erase traces of participating in deepfake exploitation.

Art direction by Samantha Amaro| Illustration by MidJourney

The Victimization of Being Deepfaked

There is shame in having your images taken and turned into something that you could not have imagined. All someone needs to be able to deepfake someone is their image — it can be taken from a yearbook photo, someone’s social media account, or people who have even used profile pictures to be able to produce this content. This not only leaves victims on edge as to how they interact with others, or how much they post on social media, but it can impact their futures. As seen in the case of a teen in Texas who was a victim of this, she describes the following “Berry and other victims of fake nudes say they worry about the long-term ramifications of such images surfacing when they apply for college or for jobs. There are immediate concerns, too. Berry says she was worried about telling her mom and stepdad.” How can you prove to someone you know that a deepfake is not real? This leaves victims feeling vulnerable if they have no one that they can confide in. Additionally, it could have long-term effects on their future if it gets into the hands of future employers by tainting the image of people they are the ones who are left to suffer while those creating deepfakes take no accountability.

No Trace Left Behind

There is also no trace to tell if the images, videos, or audio being spread is fake. There are no watermarks or indicators to know that an image is not real. According to Sensity AI, “approximately 90–95% of deepfake videos since 2018 were primarily based on non-consensual pornography.” Besides being destructive to one’s image, it is important to note that almost 90–95% of deepfakes being created are non-consensual, they are using a person’s image to create situations where victims are manipulated and left scarred. They have no way to verify where these images have come from, which also leaves the companies that allow this to happen with no accountability.

How are Deepfakes Produced?

As seen in some of these instances all one needs is a couple seconds of a video, sometimes even a profile picture, or a couple of seconds of your voice to be able to produce this content. The creation tools use a technique known as a “puppet deepfake”, because it maneuvers one’s body in certain ways. It is the “use Generative Adversarial Network (GAN) technology which consists of computer-based graphics . One of the most common deepfake techniques is face swaps “ to produce a convincing face swap involves AI. The technology allows an adversary to swap the face of one person onto another person’s face and body” The computation of this can differ, however, there are a variety of apps out there that allow for this face swap technology — from Faceshifter to DeepFace lab to TikTok and Snapchat.

Art direction by Samanta Amarao and Summarize Wise Custom GPT | Art by ai.invideo.io

Awareness of AI Use

Some of the steps that can be taken to bring awareness of deepfake technology being used not just in private, but posted on social media is to let users know that the content they are consuming is AI. For example, “TikTok said last month that it was introducing a label that users could select to show whether their videos used AI” So, instead of bluntly hiding that creators are using AI, they can choose to share that there is AI use. But there is still a con to this because it is a choice, and there is no guarantee that they will share the AI that was used in their content. So, still, we see how there is some permissibility in the use of AI even in spaces such as social media. This is bad because as we have seen, deepfakes can be spread not just in private, but publicly and often in spaces such as TikTok: “The use of A.I. voices to share mostly salacious gossip and rumors offered a road map for bad actors wanting to manipulate public opinion and share falsehoods to mass audiences online” People are being allowed to spread false information online without warning.

How Can We Raise Awareness of AI in Youth/ Adults?

Now, we must look at some precedent that has been set in terms of accountability for deepfake exploitation. It is important to note that it is not just adults who are engaging in the spread of these deepfakes but teenagers are voluntarily participating: “firstly, 80% of those apprehended by police for deepfake sex crimes are in their teens. Among them, 25% are under 14 and therefore not subject to criminal prosecution.” Because we are seeing minors sometimes even creating pornographic content of minors without prosecution, there is no repercussions being faced and there is nothing for them to learn. Instead, they are being protected and victims are left feeling defenseless. This is why it is important to discuss the importance of ethics when it comes to creating and consuming deepfake content. According to the following journal article, “An essential first step is simply to increase public awareness of the possibilities and dangers of deepfakes. An informed citizenry is a crucial defense against widespread misinformation” While it is important to create laws to protect victims, we must also preach the message as to why creating deepfakes is wrong, and the criminality of them. Because, many times they are seen as pranks and not something that could ruin one’s life. And the lack of intervention by schools, with the loopholes in the legal system that allow for these situations to continue bubbling underneath the surface.

Legal Precedents

Moreover, “law enforcement officials can rely on existing criminal laws until deepface-specific federal laws are passed. In some cases, victims can pursue a civil suit by claiming extortion, harassment, copyright infringement (i.e. using images protected by copyright law without permission), intentional infliction of emotional distress, defamation, false advertisement, and false light (i.e. a form of invasion of privacy)” By using pre-existing civil suits and clauses at least in the US, countries such as South Korea who have seen an upshot in deepfakes can work with law enforcement to create protection for victims and repercussions for those who create deepfakes.

Current Legal State of Deepfakes

In “several U.S. states and many other countries have passed legislation to criminalize the production, sale or possession of fabricated media, but these are usually isolated to specific purposes, like political and sexual content (especially the sexual depiction of minors)” The current system seeks to protect minors and reprimand exploitation against them, but we must also look to the women, who range from mothers and teachers, as well as celebrities, who are also facing victimization. Currently in South Korea “a proposed “deepfake bill” to revise the current Act on Sexual Crimes of Violence, so that it can set a penalty of up to seven years in prison for those who distribute fake videos” It can also take a long process to get bills approved to allow penalties for those who distribute these fake videos, which will in the meantime allow for this content to keep being created.

Deepfake Technology or Human Rights?

The elimination or censorship of generative AI calls into question what freedom of speech individuals have when sharing AI content online. But there are too many loopholes that allow people to spread deepfakes like wildfire, so it should not just be a matter of more legal precedents to be set, but rather these platforms should also do more to protect individuals. Some questions they can ask themselves when creating deepfakes or sharing them can be as follows: “ Can this software or app be used to impersonate someone? (e.g., does it mimic someone’s voice? Does it fake someone’s ID?) − Is this software or app safe for users, that is, it protects user’s privacy? − Is this software or app used to create porn videos from synthetic data? − Is this software or app used to create fake news? (e.g., does it fake a politician’s speech video?)” If they answer yes to any of these questions the credibility of these platforms needs to be checked, and there should be more put into place to allow people to tell that it is in fact fake content that is being spread.

Concluding Thoughts

Overall, while the use of AI has increased, it has gone further than just asking generative AI to create art or outlines — it is being used more maliciously to create deepfakes. And while I do not think that one’s freedom of expression should be limited, there are loopholes that allow perpetrators of deepfakes to walk away with no consequences. There is also the complicitness of apps for the role they play in the spread of not just pornography but child pornography, and the accountability they must take. While some of these creators’ intentions in mimicking voices or faces are not malicious, there are people with malicious intent using AI technology to create exploitive content. There are issues concerning deepfakes being observed not just in the United States, but across the world such as in South Korea. More needs to be done to protect women and minors from this. Deepfake technology is not fun, and it questions the notion of one’s image, and what protection one has in simply existing. There needs to be more regulation of deepfake technology and more penalties for the use of it.

Art direction by Samantha Amaro | Illustration by MidJourney
Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *