Madhumita Murgia On Why Women Are Unfairly Plagued By Deepfakes

As the lines between reality and the virtual realm continue to blur, a contemporary threat to online safety has emerged – the rising popularity of deepfakes. This technology that facilitates the fabrication and manipulation of identities disproportionately targets women, making them vulnerable to exploitation and sexual harassment. Deepfakes not only undermine the integrity of individual identity but also have serious implications for society's relationship with the media. In the latest episode of The Rulebreaker Show, Madhumita Murgia, the author of Code Dependent: Living in the Shadow of AI, analyses the perils that today’s digital landscape poses for the world.

Speaking to Shaili Chopra, founder of SheThePeople and Gytree, Murgia pointed out, "About 99% of deepfakes online are pornographic and about 98% of that content is deepfake pornography of women." Delving more into this alarming trend, the writer and AI expert highlighted the importance of monitoring such technology which is headed towards a dark path.

Why Deepfakes Are A Threat To Women

Madhumita Murgia, author and AI savant spoke about the multifaceted dangers of deepfakes, particularly to women. "Women are extremely targeted by deepfake technology. If we don't regulate on a global scale, we would be walking into a problem that's only going to get worse," she pointed out.

When asked whether artificial intelligence is gendered, Murgia vehemently argued that any synthetic media is inherently marred by gender bias. "The engineers and scientists who build these systems make ethical decisions and set boundaries for the outputs of the system. So who they are is really reflected in the building of AI technology," she stated.

Call For Gender Diversity In AI

As Madhumita Murgia said, most scientific advancements that we have today are built on the foundations of human bias. For example, Saudi Arabia recently unveiled a 'female' humanoid who refuses to speak about politics or sex, owing to the country's strict laws. The engineers behind the robot programmed it to align with gender roles in their culture.

Murgia stated, "You can see examples where there are recruitment algorithms where AI tends to be biased towards previous people who have been employed by a company. Or if you look at generative AI, images of women are often hypersexualised. These technologies today are gendered but I believe that there is a focus on trying to improve that."

The biases mentioned by Murgia emphasise why there needs to be more women on decision-making platforms and boardrooms of the tech industry. Equitable representation of women from different communities and backgrounds not only proposes opportunities but also potentially avoids output errors like sexism or racism from AI technology. 

Solution To The Deepfake Problem?

In a world where social media validation is slowly integrating itself into our lives, like a new addition to Maslow's hierarchy of needs, our privacy is also slowly eroding. In this digitally fixated world, the challenges that women once faced in person have also permeated into the online realm in the face of cyberstalking, cyberbullying, online sexual harassment, or revenge porn.

So how can women be safe from deepfakes?

Murgia expressed, "There are thousands of ordinary women-- not celebrities, not asking for attention-- whose images are deepfaked. (They) are not asking for it. It is in no way (their) fault. The way for women to be protected is not by going off social media or hiding from the world, but there has to be some protection from governments to ensure that people who are twisting the technology and wielding it as a weapon against women are held accountable."

The proliferation of deepfake technology poses a significant threat to women's autonomy and safety in the digital age. As Madhumita Murgia suggested, these problems not only necessitate governmental intervention but also a space for more female representation in tech, so that women can provide solutions from their own experiences and expertise.