This interview was conducted by Ula Rutkowska (Senior Researcher, Apolitical) and edited by Christina Obolenskaya (MSc in International History, LSE and Communications Intern, Apolitical). Jimena Sofía Viveros Álvarez was named on Apolitical’s Government AI 100 2025.

Global AI governance often fails to account for the realities of communities in the Global South. AI adoption cannot look the same in Mexico, Zimbabwe or Vanuatu, yet many frameworks assume a one-size-fits-all approach.

This conversation explores the disconnect between high-level AI discussions and the Global South’s needs, the risks of deepening inequality and how AI should be leveraged for peacebuilding rather than conflict. It also examines the role of grassroots initiatives in ensuring AI is responsible, secure and inclusive, particularly for marginalised communities and those most vulnerable to its risks.

Q: Could you start by explaining the disconnect between global AI governance discussions and local realities? How do current conversations fail to address local concerns?

What I am normally asked everywhere is, "Great, you're talking about safe AI, but what does that mean for me? What are my obligations, my rights? How do I implement this? How do I operate it safely and transparently and how do I protect myself and my community?" These abstract principles within high-level governance are meant to be truly global, especially in the current landscape. But the translation of that varies. In Switzerland, it might be clear-cut, but in Mexico, Zimbabwe or Vanuatu, it looks very different depending on the context and capabilities.

Everyone is focused on AI, quantum, and other emerging technologies impacting our world, but they forget large segments of our population do not even have electricity. How do you reconcile that? We are talking about digital public infrastructure and common goods, but we are not starting from the same baseline. This factual reality is not at all well integrated into global governance approaches and benchmarks for measuring implementation. We are allegedly trying to inject accelerators to narrow these differences yet they are separating us further, and sadly the Global South is generally used as a token rather than a genuine governance priority.

This is why I have focused on a “grassroots approach” to responsible, safe, secure and trustworthy AI and innovation. Through my consultancy firm, iQuilibriumAI, we tailor solutions for various clients with great confidentiality, especially in peace, security, humanitarian action and environmental protection. At the same time, I run the HumAIne Foundation, a non-profit that tackles the same issues to ensure broader meaningful impact for those who cannot afford paid services.

Within this Foundation, we have created the Global South Synergies and Resilience Platform, which connects start-ups, entrepreneurs, students, professors, researchers and academics to foster technological resilience. It also creates synergies to connect those who are remote and far away from traditional resources, and share best practices amongst all of us in the Global South for our common good and development. The goal is to foster innovation and stop “reinventing the wheel” everywhere all the time, which in turn leads to more peaceful societies. Unfortunately, the Global South remains largely a consumer of AI technologies rather than a producer, making us more vulnerable to risks. Our aim is to level the playing field with a multi-stakeholder engagement approach, amplifying and echoing voices that would not normally be heard.

We have a strong focus and commitment to women, ensuring that 60% of everything we do benefits women. We also prioritise vulnerable and marginalised communities, including indigenous communities, people living with disabilities and senior citizens. My motivation is to bridge the gap between high-level governance discussions and the tangible needs of the realities on the ground, ensuring that AI does not enhance vulnerabilities but instead empowers Global South communities to build our own technological resilience for and by ourselves.

Q: Could you walk me through some of the equality concerns around AI and how it is exacerbating inequality?

The inequalities stem from historically systemic issues but are exacerbated by AI’s replication of bias and discrimination. Many AI systems are trained on data that reflects societal biases, leading to decisions that exponentially reinforce existing inequalities rather than mitigate them.

This is especially concerning in contexts where regulatory frameworks are weak, resources are limited and local community voices are excluded from global AI governance discussions. Without an inclusive contextualised approach, AI greatly risks deepening divides rather than closing gaps. That is why our work focuses on a community-based approach — bridging links, fostering South-to-South collaboration and ensuring that AI development and deployment align with the needs of those most affected by its risks.

Could you go into greater detail about your vision for safe AI?

This answer is largely dependent on the domain and specific use cases. My main area of concern is peace and security. As UN Deputy Secretary-General Amina Mohammed rightly stated, “We cannot have sustainable development without sustaining peace, nor (sic.) can we build a secure future for everyone without addressing the root causes of our conflicts and vulnerabilities.” Therefore, I maintain it is crucial to address structural disequilibria that create and perpetuate conflict. AI is being used to make weapon systems and decision support systems that are deployed by all sorts of actors and are field-tested without complying with international humanitarian law, human rights or even technical safety standards. This is extremely concerning to me. Safe AI should emphasise the peace and security domain, which is often wrongly excluded from global conversations due to silos and arguments about the veiled nature of national security.

This is where the highest stakes and most human rights violations occur with impunity — and no accountability or recourse for remedy.

Q: Building on that, what does it look like to promote responsible AI use in a military context and where do you see the greatest need for improvement?

First, we need to recognise that any global governance of AI needs to acknowledge that the technology is intrinsically dual-use by nature. It is absolutely key to acknowledge this dual-use for the baseline principles of safe, responsible and trustworthy AI development. I cannot stress this enough. Establishing exact safeguards for these technologies through their entire lifecycle — from sourcing to deployment to decommissioning and everything in between — is crucial due to the significant consequences.

Q: Do you see any reasons to be optimistic about AI potentially supporting peacebuilding or preventing conflict?

If we use AI and other emerging technologies to address existing structural or systemic inequalities, it will naturally reduce the incidence of conflict. Conflict arises where there is an absence of a legal system and weak institutions because the non-existent rule of law leads to human rights violations. On the other hand, using AI for everyone's benefit to create a better social environment and innovation ecosystem to help simultaneously resolve our most basic needs and our most pressing challenges, can meaningfully reduce conflict and make peacebuilding, peacemaking and peacekeeping efforts more successful. We should focus on these positives rather than promoting AI to destabilise societies, deepen and entrench structural inequalities or facilitate the waging of war.