Nude Deepfakes and Bias: The Rise of Feminist AI

Global - Ekhbary News Agency

Nude Deepfakes and Bias: The Rise of Feminist AI

In an era of increasing reliance on Artificial Intelligence, issues of gender bias and non-consensual nude deepfakes are emerging as serious threats to digital equality and personal security. While technologies advance at an astonishing pace, their fundamental flaws are also being exposed, rooted in the data these systems are trained on – data that often reflects societies marked by inequality. This global challenge is driving the emergence of crucial technological alternatives, such as the Feminist AI Network of Latin America and the Caribbean, which seeks to reshape the future of AI.

Numerous examples in technological literature demonstrate the depth of gender biases. Image recognition systems have struggled to accurately identify women, particularly women of color, leading to misidentifications with severe consequences, especially in law enforcement contexts. Voice assistants have long exclusively used female voices, reinforcing the stereotype that women are suited for service roles. In image generation, AI systems tend to associate the term 'CEO' with men, while producing female images when searching for 'assistant'. These examples are not merely technical errors; they reflect and perpetuate existing societal biases.

Estimates released by the World Economic Forum indicate that women constitute less than a third of the AI workforce globally, directly contributing to the marginalization of their perspectives in the development of these technologies. Ivana Bartoletti, an international expert in AI governance and author of a Council of Europe study on AI and Gender, explained to Ekhbary News Agency: "Artificial intelligence feeds on data that is not neutral; it reflects societies marked by historical inequalities and power relations. If a company wants fair results, it must interrogate its datasets, verify their representativeness, and actively intervene when they are not. Equity does not emerge on its own; it must be designed."

For Bartoletti, the recent case of Grok – Elon Musk's AI system that allowed for the generation of fake nude images of women and minors, a feature later suspended – "shows what happens when the security and rights of women are not considered in system design." She adds: "If tools exist to digitally undress women, they will be used. Nude deepfakes are a form of humiliation and control. The implicit message is dangerous: you are online, so you deserve it. This is how many women are silenced and abandon the digital space."

It is in this context that technological alternatives are forming to rethink Artificial Intelligence and transform it into a ground for struggle and shared power. For example, the Feminist AI Network in Latin America and the Caribbean supports dozens of projects oriented towards transparency and public policies. There are tools like AymurAI, Arvage IA, and SofIA that apply a gender perspective to legal analysis and expose algorithmic 'biases' and discrimination. Afro-feminism also claims AI as a space for self-determination, with assistants like AfroféminasGpt, trained on Black knowledge and voices.

Bartoletti emphasizes: "These initiatives demonstrate that we can organize to use AI for the common good, collectively share data, and develop solutions centered on real needs. But the central point remains power. The feminist issue in AI is a matter of power: women must have more of it. Not on the margins, but at the top of companies and in places where technological policy is decided. Diversity is needed in decision-making spaces, not just among those who write code. Artificial intelligence is not just technology; it is a choice about how we want to transform society."

Ekhbary News Agency