A video created with artificial intelligence (AI) went viral on social media in July 2025, promoted by accounts affiliated with the party of Argentine President Javier Milei. It featured fabricated images of Argentine journalist Julia Mengolini and her brother, depicting an alleged incestuous relationship. She filed a lawsuit and recounted the impact this violence had and continues to have on her life. These deepfakes with violent intentions are spreading now more than ever with the use of AI.
Since last year, I’ve been wondering what we’re doing about it. I started a survey, which I keep updating, on how feminist activists, professionals, and organizations in Latin America are using AI. There’s violence, but there’s also action. Here are some examples.
First, let’s provide some context. I asked Graciela Natansohn, a researcher in feminism and communication at the Federal University of Bahia, Brazil, what feminist AI is. “It doesn’t mean AI by and for women, although in many cases that can be desirable and useful, but it’s something bigger. Feminist AI for colonized territories like Latin America and the Caribbean is a technology that protects and promotes local ways of life, that respects the time and territory of the communities it serves, that includes the perspectives of everyone in the decisions made, and that serves no other purpose than those determined by that community,” she replied.
Mailén García, director of DataGénero, also contributed: “We understand feminist AI as a political, technical and ethical approach that seeks to profoundly transform the way in which AI systems are produced, implemented and governed.”
We know that data and algorithms are not neutral, which is why García emphasized that feminist AI “is not limited to correcting obvious biases, but proposes to review and reimagine the entire life cycle of technology so that it contributes to social justice, gender equality, and environmental sustainability.”
Entangled in AI
A characteristic of Latin American feminism, historically, is its networking. Following this tradition, the newly formed Feminist AI Network in Latin America and the Caribbean is underway. Its first action was to convene those developing projects in the region. And it was a success. They received 129 proposals, 50 of which were selected. Of these, 25 are in the seed stage, 19 are prototype, six are in the pilot stage, and four moved from prototype to seed stage. Finally, 10 projects will be selected to continue their development.
The network is supported by the International Development Research Centre (IDRC) of Canada, which had already promoted other experiences between 2021 and 2024, and from where this new regional initiative arises.
Feminist AI “promotes a public purpose for technology: it must expand rights, improve policies, strengthen institutions, and produce knowledge that benefits society. This vision translates into projects like AymurAI, aimed at opening up information, strengthening access to justice, and producing evidence to transform public policies,” explained García, who is one of the members of the network and one of the creators of the software that anonymizes documents, collects, and makes available data from court rulings.
There are two more experiences in Argentina in the judicial field, but they take place within the ChatGPT field. One is Arvage AI, developed by Judge Rita Custet to apply a gender perspective to legal analysis, and the other is SofIA, by lawyer Ana Correa, who presents it as a tool that “detects gender biases and proposes alternatives that take into account the history and data involving women and LGBTQ+ people.”
From Brazil, Natansohn adds to this journey of Latin American feminist AI Nhandeflix (which means “our flix”) which operates on its own intranet and is a multimedia application developed by Guarani communities in São Paulo, which is accessible by Wi-Fi in the communities where — in addition to a messaging service — movies, audios and all kinds of files are made available “preserving and strengthening the Guarani language, culture, and spirituality”.
The AzMina Institute, for its part, created QuitérIA, an AI that “monitors legislative proposals in the Brazilian Congress, in relation to the rights of children, women, and LGBTQ+ people.”
“Avoiding data extraction and the use of data centers is a feminist principle in AI, although there is a diversity of perspectives on what might or might not be considered feminist,” the researcher shared from Bahía Blanca. “This is the case with the Plinq platform, which allows women to check if the men they are dating have criminal records or histories of violence. This is done with public data, which limits its reach (stealthy processes go undetected), and, furthermore, the service is paid and does not use open-source and auditable software. It is a start-up, an individual commercial venture.”
There’s another project I’m familiar with because I’m involved in it: Narratives of Femicide. It’s a participatory research project where we read and annotate news stories published in Latin America to create a repository of harmful and constructive journalistic practices in femicide coverage. With this dataset, the idea is to design potential AI tools to support feminist and contextualized news production. It’s an initiative of Data Against Femicide in collaboration with the Data in Society Collective (DISCO Lab) at Brown University in the United States.
Regarding gender-based violence, there are AI-powered initiatives such as Sof+IA from Chile, which identifies as a Feminist Listening System that provides guidance in situations of abuse, and OlimpIA, a platform that offers free support, guaranteeing anonymity to women who are victims of online harassment. It is named after Olimpia Coral Melo, a driving force behind legislation throughout the region to prevent and punish digital violence. Notably, it was developed by survivors.
In the regulatory sphere, the Committee of Experts of the Follow-up Mechanism to the Belém do Pará Convention (MESECVI) of the Organization of American States (OAS), recently presented in Brazil the Inter-American Model Law to confront digital violence against women, which has articles related to AI.
Democratic governance of AI is another central focus for feminists, “where decisions are made with participation, based on criteria of social and environmental justice, and with regulations that distribute power equitably. The ethics of care accompany the entire process: protecting privacy, ensuring digital security, reducing risks for vulnerable communities, and minimizing the ecological impact of technological infrastructures,” García summarized.
And she emphasized: “It’s not AI for AI’s sake. That is, we also consider when it makes sense, what benefits it will bring, and what drawbacks.” The process is underway, and it’s important to support them.
Sign up for our weekly newsletter to get more English-language news coverage from EL PAÍS USA Edition
