Innovation always outpaces regulations. And thus arises disruptive changes. Nowadays, nonconsensual deepfake pornography is surfing around the internet, making women subject to character assassination and threats. What can we do about it? What policy challenges does it introduce?
Toya, an undergraduate student, is an entrepreneur. As a part of her page’s marketing campaign, she regularly appears on Facebook live and recorded videos. One evening, a friend of hers warns her that an adult video of her has gone viral. Horrified, Toya freezes completely, shutting away from social media, public gatherings, and even university lectures. She doesn’t understand what to do, who to contact, or how to defend herself; because she’s never done any adult videos, ever!
Although the above scenario is hypothetical, the reality isn’t further from this. Just a few days ago, a popular content creator and entrepreneur has been a victim of “Deepfake Porn”, where her face has been edited into the body of an adult actress, portraying her in a sexualized context. In a tearful interview with a major daily, she expressed her agony caused by the public humiliation and demanded justice.
“Deepfake Porn” is a certain type of video manipulation technique using Artificial Intelligence (AI) that produces hyperrealistic pornographic images and videos of generally nonconsenting women using a real woman’s face edited onto the body of an adult actress. In recent days, deepfake porn has been creating suffering in the lives of hundreds of women daily, and the number is only going to increase in the future.
Uses of Deepfake
High-end video manipulation has been used for a long time in big-budget entertainment mediums. For example, after Paul Walker died in a tragic car crash, his face was digitally swapped into his brother’s face in the movie ‘Fast and Furious 7’. Recently, viewers of the Oscar-nominated 2019 film ‘The Irishman’, saw the utilization of Deepfake in ‘de-aging’ Robert De Niro to fit the role.
Deep fakes, if used properly, can allow viewers to enjoy their favorite characters and actors on screen even if the actors are too old to act, or even after their eventual pass-away.
As deepfake software are becoming more available, so is the misuse of them. Recently, a TikTok went viral where popular podcast host Joe Rogan and a Stanford facility were seen to promote a supplement, whereas, in actuality, they never promoted that particular product. Another scammer used a cheap and available Deepfake software to fake a bank official’s voice to steal 35 Million USD from a bank.
As Deepfake software improves, the misuse of them will only grow in number and severity.
Without a doubt, the biggest victim of Deepfake-related crimes is the women. According to a 2020 report by Sensity AI, 90 to 95 percent of malicious use of Deepfake are pornographic. Of them, 90% of victims are females. By October 2020, a telegram Deepnude bot alone targeted over 100 thousand nonconsenting women
The primary victim among them is female activists, actresses, and politicians. Nina Jankowicz, disinformation researcher and author of ‘How to be a Woman Online: Surviving Abuse and Harassment, and How to Fight Back‘ (2022) was a subject of deepfake porn herself, a very phenomenon she is fighting against. The same fate was shared by the founder of ‘Not Your Porn’, a group fighting to remove nonconsensual porn from the internet. Greta Thunberg, an environmental activist, and Hillary Clinton, a former US presidential candidate were subject to Deepfake porn as well.
Nina Jankowicz
Greta Thunberg
Hillary Clinton
But the most alarming news is, deepfake technologies have become so good that anyone’s face can be used to produce deepfake porn with a handful of images and a couple of videos. Not surprisingly, making deepfake porn with anyone’s face will become even easier as the software gets better.
How Deepfake Porn Harms Women
In the medieval era, women’s faces were drawn on posters and leaflets, portraying the accused women in profane acts with the devil before declaring them as witches. Then the drawings were disseminated to humiliate the accused in public eyes before burning them alive (commonly known as witch-hunting). Needless to say, all of the drawings were fake and nonconsensual.
Similarly, the reason behind subjecting an ‘everyday woman’ to deepfake porn is primarily public humiliation. Unfortunately, although society has gone through a lot of struggle to embrace modern ideas, somehow it could not shed the patriarchal ideas of women’s ‘purity’ and ‘chastity’.
Remember Toya from our hypothetical scenario? Just like the scenario, just being subject to a deepfake porn brings an unimaginable amount of tangible and intangible suffering to a woman.
Employment Issues
Firstly, being a victim of deepfake porn would almost guarantee that no employer would hire that victim in his company. Even if a female victim is self-employed, she would definitely find it difficult to find business partners. The suffering is even worse for online-based entrepreneurs and content creators who already go through heaps of harassment just for being women.
Harm to Family Members
Secondly, the embarrassment would also harm the family members, specifically the children who cannot easily deal with bullying and harassment. For example, after a deepfake porn went viral of a Twitch streamer named QT Cinderella, someone anonymously sent the videos to her underage cousin as a form of online bullying.
Romantic Partners
Thirdly, just being a victim would make it incredibly difficult for a woman to find a romantic partner because of the taboo and stigma attached to it.
Online Harassment
Fourthly, being online or in public places would be incredibly difficult for a victim afterward because of the immense amount of trauma a female would face. Even for professional adult stars, who understand what’s about to come with their job, it is still hard to cope with the constant harassment, judgment, and berating. But for those who didn’t even sign up for it, it is hard to even imagine what they would have to go through.
Harm to Other Women
Fifthly, even if a female isn’t a subject of deepfake porn, deepfake porn is still harmful to her. There have already been numerous accounts of women who found that even their nonsexual pictures are being sexualized by AI because that is how AI has been used primarily to modify women’s pictures.
Last but not least, even children are not safe from deepfake porn. A 20-year-old McArthur Highschool student was sent to prison after being guilty of producing deepfake porn using children’s faces.
What Women Can Do To Protect Themselves from Deepfake Porn
From a personal responsibility standpoint, women can be more careful about their sharing of pictures and videos on social media. Locking Facebook and Instagram profiles, making pictures unavailable for download, and not accepting requests from strangers can be the initial steps. In the case of deepfake victims, contacting Police Cyber Support for Women, and submitting complaints on Google’s complaint page should be the immediate steps.
But needless to say, females’ being paranoid about using social media is not the intended outcome here. Everyone has the right to use the internet without any fear of harassment and judgment. One’s likelihood is one’s own property, and misusing it in any way by another person is a forgery which is a punishable offense. Misusing one’s likelihood in any sexual way without consent is outright sexual harassment and is a heinous crime. Thus, government and international organizations must come forward to protect women as soon as possible.
What The Government of Bangladesh Can Do To Save The Women
Innovation always outpaces regulations. But in a country like Bangladesh, where socioeconomic condition already puts women at a massive disadvantage; a massive threat like deepfake porn should be regulated preemptively. There are several steps Bangladesh can take drawing from regulations of other countries.
Curbing the Distribution
Social media and video-sharing sites like Facebook, Twitter and YouTube have already put policies to ban deepfake porn immediately. Other sites like Reddit and Telegram have also followed suit.
But only curbing the distribution is not enough. Firstly, the huge increase in the publication of deepfake porn makes it harder to control their distribution of them. Secondly, there are a lot of separate sites specifically built for sharing deepfake porn which can bypass the regulations of any social media sites. For instance, Genevieve Oh, a researcher who studies livestream, found that from 2018 to 2022, the number of deepfake porn published on the top site dedicated to Deepfake porn has ballooned from two thousand to thirteen thousand.
As the AI gets better, it is predicted that curbing the distribution would be even harder because the fake would be almost indistinguishable from the reality.
Watermarking
A second viable way can be creating regulations to force deepfake software companies to watermark fake videos. Both China and the EU have passed policies regarding this.
However, although watermarking is a necessary step, it is not sufficient. Firstly, just by a video being published, the damage can be done. Going back to our hypothetical scenario again, Toya’s life would still be upside down even if everyone around her knew that the video they saw was fake. Secondly, even if the major deepfake software like Dall-E and Midjourney has policies prohibiting making adult deepfakes with real faces, due to consumer demand, other software companies may allow their users to do so to gain an advantage over Dall-E or Midjourney.
Penalizing Deepfake Porn
To deter people from making deepfake porn in the first place, creating laws penalizing it may be needed. England, Wales and Australia are planning laws banning the creation of nonconsensual deepfake porn.
However, even penalizing it may not be enough. Most of the perpetrators lie behind the curtain and thus it becomes really difficult to bring the criminals to justice.
Banning Problematic Software and Sites
The most viable way of deterring the production of deepfake porn is banning deepfake distributing sites and noncompliant software altogether. By restricting access to
- the noncompliant software that does not prohibit the production of deepfake porn, and
- sites that are associated with distributing deepfake porn
The government of Bangladesh can curb both the production and distribution of deepfake porn, and protect its vulnerable women.
Bangladesh has done very well in the department of women’s empowerment. Incumbent Prime Minister Sheikh Hasina has been awarded ‘Planet 50-50‘ and ‘Agent of Change‘ for her contribution to women’s empowerment. Creating effective policy tools to curb the production and distribution of deepfake porn would allow the government to show its commitment to showing zero tolerance against sexual harassment and exploitation. Only a concerted effort by all parties concerned can protect women from such a novel and impending problem.
About the Author
All-Marufi Rahaman Sajon is a senior at the Department of Development Studies, University of Dhaka.
1 comment
Excellent writing❤️