AI governance for and from the Global South

Building the basis for a sustainable and human rights centered development

Open Data Charter
6 min readDec 19, 2023


by Jamila Venturini, co-Executive Director at Derechos Digitales,

By Gilda Martini, license CC

It is true that AI could foster more just and sustainable societies. But this will depend on the conditions for its production, deployment and assessment. Like any other technology, AI cannot and will not solve historical social problems by itself. To date, what evidence increasingly shows is that AI development and implementation has already impacted the exercise of fundamental human rights–particularly within historically marginalized groups.

The AI industry more broadly, but also AI applications and users, are distributed in a widely unequal manner among and within countries, and so are its potential benefits and harms. Even the incipient AI industry in Latin America is still built from infrastructure controlled by international actors (including cloud services) and using training datasets developed elsewhere with, as we know, little to no ethical considerations.

The way the global governance of AI will be shaped is a central piece of a very complex puzzle being played today. While we advance towards developing rules to regulate AI in countries from the Global South, some considerations become of extreme relevance despite the diversity of contexts they represent.

Priorities from the Global South

On the one hand, it is necessary that the starting point for any process is a human rights perspective and countries’ existing obligations with international norms and standards.

Human rights impacts derived from AI may include a variety of violations that often could (and should) be prevented with existing measures and mechanisms, without reinventing the wheel. International human rights experts have already provided guidelines on how this can be operationalized in order to prevent further harm, and international entities including the UN Human Rights Council continue advancing standards to protect human rights from the risks of advanced and emerging digital technologies.

Generic ethical frameworks or so-called principle-based regulation, will not be effective to protect our populations. Worse than that, they risk weakening existing rights and obligations, or validating technologies that are known to be abusive. Such is the case of facial recognition systems, for instance, which are increasingly used for different purposes in the Global South.

Implemented for public security, it consists of an AI-based mass surveillance technology that transforms millions of people into potential suspects. Additionally, it already accounts for the detention of numerous innocent people, mostly individuals of racialized groups, in contexts where racism influences the disparity of policing. Used to control access to public spaces or services, it has the potential to exclude the ones it can not recognize–which according to studies consists mostly of women of color. In Brazil, they are advancing even in public primary schools to control students’ attendance, risking subjecting children and adolescents to different forms of discrimination and extracting sensitive biometric information from them without any possibility of a meaningful consent.

In these cases, and several others of AI deployment, safeguards offered to data subjects are minimum, including against future use of their information that could result in other forms of discrimination. Research developed in Latin America by Derechos Digitales shows that countries using AI for public policies fail to comply with existing access to information and data privacy standards.

On the other hand, the protection and promotion of economic, social, cultural and environmental rights should be at the center of any normative effort, making sure that no system will become a barrier for their exercise.

Beyond the respect of human rights, it is key that any initiative to promote AI is followed by mechanisms to develop local industries and prevent concentration, as highlighted by experts from the Global Partnership on Artificial Intelligence (GPAI). It is legitimate that Global South countries dedicate efforts to have a share in the global AI production, since it partly relies on the extraction of material and non-material resources from developing countries. This would be a way towards economic welfare, which is important to achieve social rights.

A superficial attempt to regulate AI that does not consider its specific applications can make it harder for victims of abuse to claim for justice or reparation. At the same time, ignoring the underlying inequalities of the AI industry and the need to counterbalance its economic power with incentives for smaller and alternative players, risks to continue benefiting the actors that already dominate the market, thus perpetuating chains of dependency and exploitation.

Strengthening commitments, overcoming silos

As it becomes evident, local norms will not resolve the issues presented at the global level. Until now, most of the AI governance debate has been guided by industry and foreign perspectives and priorities. Research from 2019 already showed how most of the main global documents on AI and ethics were led by the private sector and there existed an under-representation of Global South actors in the main discussions around AI.

Global South countries need to advance a common and operational agenda to participate in the (continually multitudinous) global conversations on AI. While some advances on conversations are observed within the African Union and Latin America, more coordination is needed to make a common voice reach other spaces, as well as a concrete commitment to advance processes that are still a debt for several countries regarding transparency and accountability from the public sector and data protection.

We need to strengthen the call for mechanisms of control that assure that human rights are respected throughout all the AI life-cycle and by all industry actors, regardless of where they are based. Beyond incentives to deploy technology that resulted in human rights abuses in other regions, we have to demand that strict limitations are respected regarding their exportation and that support is focused on building local models that are sustainable and ethical. Calls for moratoriums on certain technologies or applications should be considered seriously at the national and global levels, and be heeded by technology companies to abstain from developing rights-infringing technologies.

Meaningful participation for meaningful policies

AI development highlights an opportunity for local development that is grounded in human rights and regulation definitely has a role in fostering such type of development. However, there seems to be a disconnection between policy advances and the key challenges presented by AI both inside Global South countries and within international policy spaces. While some governments rush or are pushed into a need to legislate on AI, several underlying legislation are either absent or failing, with invasive technology being deployed far from the public eye.

To build a proper agenda to tackle the issues that actually concern Global South contexts and populations, it will be necessary to include those who are most affected by AI developments, as well as the ones who could benefit more from it and who may have relevant proposals to be considered. Until now, civil society participation has been limited and permeated by tokenism, leaving relevant perspectives and realities completely invisible, and maintaining broad participation as validation for harmful practices without meaningful, structural changes to the way in which technologies are built and used.

Thus, we need to rethink AI governance from the ground up, from its ideation to the construction of the tools and machines that make it possible, to the knowledge that is needed and shared to make it happen. Complex problems cannot be tackled effectively by targeting small portions at a time, without a broad view that takes into account the whole of the problem. A source of problems that affects everyone, can be best addressed by including those that can represent everyone.

Jamila Venturini is the co-Executive Director of Derechos Digitales, where she coordinates the organization’s efforts on artificial intelligence and inclusion in Latin America.

This article is part of our Finding the ‘Rights’ Balance blog series which was kickstarted by our Research Director, Renato Berrino Malaccorto. Part One discusses how access to public information and open data complement each other, while Part Two presents 7 ideas that harmonise debates surrounding open data and privacy.



Open Data Charter

Collaborating with governments and organisations to open up data for pay parity, climate action and combatting corruption.