Thematic discussion 2: Bridging AI divides - Global Dialogue on Artificial Intelligence Governance - Day 1

Can I ask you to take your seats, please, those who are standing? Good afternoon, everyone. Excellencies, distinguished co-chairs, distinguished guests, and colleagues. Good afternoon and welcome to this thematic breakout cluster focused on bridging AI divides. My name is Robert Opp. I'm the Chief Digital Officer and the Director of the Digital AI and Innovation Team at the United Nations Development Programme. It is a pleasure to be one of your MCs for this afternoon, together with my colleague Loretta Hébert-Girardet from UN Disaster Risk Reduction. I'm very pleased to be here today, especially in this session that is focused on an issue that is top priority for UNDP. 2026 is truly turning into a turning point for how the world responds to AI. This session is about bridging the AI divide, and there are likely many AI divides. We in particular see a divide around access, which is affordable and available access to compute power, capacity, and skills, other things. And we see a divide around adoption as well, which is where AI is already entering countries through procurement systems, through other vendors, and outstripping and outpacing the ability of institutions to keep pace. And therefore, the divide around adoption is how we ensure that AI is safely and inclusively adopted at a country level. So, in our work with more than 170 countries, UNDP sees that AI is evolving from a technology conversation to an implementation and systems challenge, underscoring the importance of capacity building and good digital foundations. Before we move forward with an amazing lineup of speakers and distinguished guests, I want to turn to my colleague Loretta for her opening remarks. Thank you very much.

Excellencies, distinguished co-chairs, distinguished guests, colleagues, ladies and gentlemen, good afternoon. So, as mentioned, my name is Loretta Hebergerade. I'm the Chief of Risk Knowledge at the United Nations Office for Disaster Risk Reduction. Together with my colleague Robert Opp, we are very pleased to support this session as the UN Interagency Working Group on AI co-leads and MCs for Thematic Cluster 2. Bridging AI divides is more than simply expanding access to AI. It is about strengthening the conditions that enable countries, communities, and stakeholders not only to use AI, but also to develop, adapt, evaluate, and govern the technology in ways that respond to their own needs and realities. From a disaster risk reduction perspective, AI is already beginning to transform how we understand, anticipate, and reduce disaster risks, from risk assessment and early warning systems to risk-informed investments and actions with direct implications for public safety and the protection of critical infrastructure and services. Now, this creates an important opportunity to accelerate progress, but it also reinforces the need for sustained capacity strengthening and technology transfer, particularly in developing countries. AI cannot substitute for strong institutions, trusted data systems, local expertise, community engagement, or accountable decision-making, nor should countries become dependent on technologies that cannot be interrogated, adapted, or governed. The priority, therefore, is not simply to deploy AI tools, but to ensure that countries have the digital infrastructure, institutional capacity, technical skills, and governance frameworks, as well as locally led solutions needed to use AI responsibly, safely, and effectively. So today's discussion is therefore both timely and necessary. The task before us is to identify what it will take to build the foundations needed for inclusive AI development, deployment, and governance across different contexts. So I look very forward to the conversations ahead and to hearing the priorities and solutions that emerge from these exchanges. And with this, I hand back over to Robert, who will invite our distinguished thematic cluster co-chairs for their opening remarks.

Thank you very much, Loretta. It is now my pleasure to invite to the podium here with us His Excellency Samba Diouf, who's the Minister of Communications, Telecommunications, and Digital, Republic of Senegal, and Mr. Yovan Courbalier, who's the Executive Director of Diplo Foundation. Minister, you have the floor when ready.

Thank you. Thank you very much. Good morning, everybody. I'm very happy to be here. I'm very honored. And it's a great honor for me to be here discussing about our exciting topics. AI today is shaping our lives. AI is shaping and driving our destiny, so we should be aligned on this. It's a great honor for me to co-chair this session alongside with Mr. Yohan, Executive Director of Diplo Foundation, on these exciting topics: Bridging the AI Divide Through Capacity Building, Equitability access and strong digital foundation. AI offers a lot of opportunities, and those opportunities can only be fully realized if there are unified access. Today, disparity in the connectivity are very broad. People who develop AI do not align with the consumer. Our collective responsibility is to ensure that the transformative technology becomes a driver of shared prosperity rather than a source of deeper inequality. We must create the conditions that enable every country, particularly the developing countries, to build its own capability, adapt AI to its national priorities, In Senegal, this vision is rather heard of our New Deal technology, which is based on the big pillar of digital skills, AI, digital infrastructure, public platform, and so on, and innovation. Our country can meet a lot of challenges. We must strengthen our collective effort and ensure that the benefit will be for all of us. I hope today's session will be— will help identify practical solutions, foster stronger partnerships, and generate ambitious recommendations for more inclusive, equitable, and development-oriented global AI. I thank you very much for your attendance, and thank you very much for your time. Thank you.

Thank you very much, Minister. Mr. Kobalia.

Thank you, thank you very much. Good afternoon. I'm really honored today to chair today's session with Minister Diouf Samba, and when I was preparing for the session, I thought that we are lucky because we have to deal with a topic which is a concern of many: education, writing essays, plagiarism, divides, cognitive and all sorts of divides. Therefore, Our discussion is, if you walk outside the Palexpo, is of concern of people in Palexpo, people in Senegal, people in Japan, Balkans, everywhere. Now, what is the underlying— in this 2-minute intro which I'm asked to make— what is the underlying point? AI is about knowledge, and I'm insisting, not data, knowledge. AI is created of knowledge of all people who lived before us, From Shakespeare, Dostoevsky, to Goethe, to great Ubuntu tradition of Africa, oral tradition. Therefore, this is the first aspect of knowledge. Second aspect of knowledge is that we are generating knowledge. Today, 500 of us in this room will generate some knowledge, insights, reflections, ideas. And this is excitement of both this point and topic that we cover. Let us finish today's session with new insights, new ideas, new knowledge created for many people, communities, citizens, and countries that are waiting for answers. From education to divides, cognitive knowledge and other divides. Therefore, topic is important. It is a great moment in time. It is a great group of people, and let us use it to the maximum to generate something new, not by AI, by human wisdom, creativity, and empathy, I would say. Thank you.

Thank you, Mr. Cavaglia. Okay, now I'm very pleased to introduce the next segment, which is a fireside chat with 2 incredibly distinguished guests. There's also a little bit of choreography involved. I'm going to introduce our fireside chatters, and then the 4 of us are going to move off of this podium, and we will invite our extremely distinguished guest to come to the stage. So with that, I am very pleased to introduce that we have for our fireside chat today His Excellency Félix Augusto Antonio Ulloa, Vice President of the Republic of El Salvador, and His Excellency— sorry, let me do your name again, Your Excellency. I apologize because I missed one of the names. Forgive me. His Excellency Félix Augusto Antonio Ulloa Garay, who is the Vice President of El Salvador, and His Excellency Dr. Khaled El-Anani, who's the Director-General of UNESCO. So with that, let's welcome them as we move off and they come up. Thank you.

Si vous permettez, je commence. Your Excellency, Mr. Vice President, Your Excellencies, it gives me a great pleasure to be with you today and to meet again. I had the pleasure of welcoming you in Paris last February and we had a very fruitful meeting once again this morning. I'm the new Director-General of UNESCO and I can tell you that more than 70% of my invitations that I'm receiving in any panel, it's on AI. UNESCO is very well known for culture, for heritage, for some aspects of education, but the positive surprise when I joined UNESCO last November was that I had 9 Assistant Directors-General. 4 of them are dedicated to science: natural science, social and human science, ocean, theoretical physics. So science is very much present at UNESCO. We look to science from the lens of human rights, that every people, every community have the right to participate, to get benefit, and to share, including AI, of course, having in mind to respect the human dignity, contributing to peace and sustainable development. Our objective is to make AI ethical, inclusive, and safe. To do this, we cannot without bridging the challenge, bridging the gap, and the gaps are many, including connectivity, infrastructure, skills, and legal frameworks. If you allow me, Your Excellency, I'll be sharing with you in a few minutes what UNESCO is doing now with AI. First, we are a global standard setter. Our 194 member states adopted in 2021 the first worldwide recommendation on ethics of AI. And now, it's our turn to turn from principles to implementation. We are working with governments, with the private sector, with civil society to implement. Maybe I'm coming back to your very successful experience in your country. Capacity building, which is a very important topic that we cover. During the last few years, we organized more than— training programs and tools for more than 50,000 civil servants, and the judicial actors all over the world in 192 countries. Inclusion and equality are very important for us. Everyone should benefit from AI, particularly women, youth, Africa, and small island developing states. Now, with my colleagues in the communication information sector, we are thinking launching soon a new instrument, normative instrument, on youth in the era of AI, how we can use in a very positive way this social media, this online gaming, and the AI with our teenagers and youth. And by the way, recently, a couple of weeks ago, we published a very small booklet at UNESCO with our friends at CLIMIE in France. It's a guide for parents. How to accompany the teenagers and the youth on social media and in the AI era. UNESCO also has a very, very big network across the globe, about more than 1,140 UNESCO chairs, 142 Category II centers, experts, academia, civil society, and a big number of private sector partners. Through the Spark AI Alliance, we connect 53 schools of public administration and partner institutions in 116 countries. And we are very happy to serve with ITU Secretariat for this very important dialogue that we launched today. Education for us is the cornerstone, and today I'm coming from Paris with a delegation from UNESCO, from the education sector, the basic science, the social and human science, communication, information, because as I said in the beginning, AI is everywhere in our daily life today. Even when you speak about the future of education, it's AI. Culture, it's AI and culture. Education is the cornerstone. We believe that every country has the right to have well-prepared teachers, curricula, institutions, and skills. Our focus is on STEM, and we opened— we launched 2 months ago in Shanghai, China, our iSTEM. It's the International Institute for STEM Education with the priority on women and Global South. Our first priority are teachers. In 2030, the world will need 44 million teachers, so we hope that AI can complement the work of teachers, not replace the teacher. We developed AI competency frameworks for teachers and students already used in more than 60 countries, and now we are moving from guidance to scale with the Varki Foundation. We will train more than 1 million teachers by 2028. Another priority for UNESCO is AI and digital literacy. We are supporting 40 countries integrating media and information literacy in their curricula. We work also with content creators to promote ethical use of AI and strengthen critical thinking online. Today, it was mentioned the cultural bias and the languages. This is a priority for UNESCO because only very few dominant languages dominate the cultural reference, excluding many communities. We promote many projects, including initiatives such as the English-Kiswahili AI Dictionary. Tomorrow, we are celebrating at UNESCO the Kiswahili Language Day. Turning to you now, Your Excellence, Mr. Vice President, can you share with us some of the flagship AI initiatives in your country that are delivering tangible benefits? For your citizens, and that may serve as an inspiration for others around the world. I know that you had created an agency. You have your legal new law, and recently Ram report. But maybe I keep Ram for another part of the question. Thank you.

Thank you, Doctor. I'm sorry about that. We used to speak in French. Mais si vous me permettez, dans cette occasion, je voudrais parler. But if you allow me, on this occasion, I'll speak in Spanish. Why? Because you don't have a form for the translation. J'espère. Désolé, mais nous sommes habitués à parler. I'm sorry, but with Mr. Director General, I'm used to speaking in French, and he surprised me because he spoke in English. But I will be speaking in Spanish. Le plus simple, c'est parler en arabe. The simplest would be if you speak in Arabic. Arabic and Spanish are two official languages of the United Nations, so no problem there. Thank you very much. To go on to the topic that the Director General of UNESCO raised, I'd like to start by saying that under this government of President Bukele. For the first time in the history of El Salvador, we are investing some 5% of GDP in education. It's our largest budget item in the national budget. And this support to education led us to reflect on the use of AI in our country's transformation. This was a commitment of President Bukele. To enter into this revolution, as he called it in his first statement at the United Nations in New York in 2019, in which El Salvador was going to opt for integrating itself into the digital era, and AI was one of the most important vehicles to be used for that. So we started by have signing an agreement with Google Cloud to be able to use Google Classroom, which is part of the Google Marketplace, Google Marketplace Plus that we're using now, in order to educate all students in the public education system in El Salvador. And in order to do that, we distributed over 1.2 million electronic devices, laptops and tablets, to students throughout El Salvador. Students and teachers are using Google Classroom on a daily basis in our country's classrooms. And then last year, in an agreement with Elon Musk, President Bukele managed to agree the free provision of the use of xAI, this platform that pertains to Elon Musk. We were able to use this platform For free, for personalization for each student. There is an AI tutor for each student within the public education system from El Salvador. This is positioning us as a country that is preparing its human talent so that new generations can be competitive in this modern world, in this contemporary world, and in this world— Thank you. in its digital age. The use of AI, therefore, in education in El Salvador is a daily practice, but this is also true in health. In November last year, President Bukele announced the program Doctor SV. It's a telemedicine program by which all of the citizens of El Salvador will have access to a doctor 24 hours a day through this system, from their telephone, from their laptop. They can consult with a doctor immediately at any time of day without having to go to a hospital. We have had 1.6 million consultations in this first 6 months. We are therefore expanding out hospitals. We would have needed thousands of doctors and hundreds of hospitals to reach that many people, but we have been able to do that through an AI platform. Recently, in a press conference just a month, a couple of months ago, the president also announced that we are now going to enter into a new phase in healthcare in El Salvador, treating more complex diseases such as diabetes and other illnesses, allowing patients to consult with doctors, doctors issuing prescriptions. They can then go to pharmacies, a network of pharmacies in the system delivering that medicine for free. And if laboratory tests are needed, these are also done for free at that stage. The doctor then reviews and refers to a specialist or not. This is AI applied to health. The health of citizens of El Salvador is guaranteed by AI, which is then saving lives. Yes. And we can say that making use of AI, therefore, is leading us to save a lot of lives of El Salvadorans, as well as educating the new generation and preparing them for the new era. But we cannot implement this public policy in a disparate way. And so we established the National AI agency. There is a whole normative ecosystem in El Salvador based on our constitution and a normative corpus which consists of 4 laws: the AI and Technology Promotion Act, which established the National Artificial Intelligence Agency; the Data Protection Act; the Cybersecurity Act; and the Act Promoting Robotics Technologies. These are 4 laws that drew from our founding principles which inspired the republic, and this inspires our implementation of AI in our country. Responsible innovation, competitiveness, transparency, human oversight, non-exclusion. Participation of the private sector. These are some of our founding principles on which we have built our AI policy with a series of instruments from AI laboratories, training centers, academia, universities, sandboxes, through to a series of instruments applied across our institutions. That are leading the way in AI. And this has an impact on the whole institutional framework— Ministry of Health, of Education, of Agriculture, because we are using this in agriculture, justice also. I was speaking today with the doctor too about this possibility to start a school for training of judges. Because justice and the justice system are also moving toward the application of AI for extra certainty and security in hearings. The public prosecutor, all of the institutions in our government that are making use of AI now. So this is why I wanted to share our current status and how far we have managed to go with AI in El Salvador. To date, I could continue to speak about other areas in which the digital— with the digital economy using AI. El Salvador, since 2021, has been positioning itself as the first country of the world to adopt a digital currency, Bitcoin, as legal tender. We are One of the first countries in the world that has an agency which certifies and authorizes enterprises working with digital assets. 113 enterprises throughout the world have established themselves in our country, but that would be a topic for another discussion going away from that. Thank you.

Monsieur le Vice-président. So, Chair, now Vice-president, I can only echo my congratulations to you, to your civil society, to the administration of your government for this transformation, this digital transformation of the country. AI, which is genuinely serving the people. This is an incarnation of exactly how AI can serve local populations. And I know that my colleagues with your teams have worked hand in hand closely over the last few months in order to launch the RAM report. This is a report which is a 2-step report. It has to do with the local authorities and to assess to what extent they're ready for the rollout, the implementation of ethical principles on AI, on infrastructure competencies, laws. And also identifying the priorities for the country and how they can move forward. And at this time, there are some 80 different countries throughout the world who are working with UNESCO. Either they've either finished the process or they're in the process, they're finishing it. And so I think on the 18th of June last, your country published this report, the RAM Report. Could you please share with us your experience and if there are any lessons to be learned for the future? Go ahead, sir.

With pleasure. For those who don't know the RAM concept, RAM stands for Readiness Assessment Methodology. That means that this is a methodology to assess the situation and UNESCO together with other partners assess the status of readiness. Puts to the country. And I should highlight that one of the sub-indicators in legislation is where we stand out in score compared to other countries in the region. And some of the giants in the industry such as Google have decided to set up a hub in El Salvador in this area in recognition of that. But there are also weaknesses with every strength comes weaknesses, and these have also been flagged up by the Ram Report. One is the gap between the use of these tools in rural and in urban areas, and we really do have to ramp up work in rural areas. We saw this when we started to roll out Google Classroom in schools. El Salvador is a small country in terms of landmass, but there are mountainous areas where connectivity is low. And we were seeing with students, we'd give a laptop and they'd start to play games because they couldn't connect. They started to play solitaire. So we tried to find solutions to this. And one was White Space, a kind of router that the local Governments were— the town halls were contracting in order to provide coverage in a limited area. This was a limited solution. Then we saw President Bukele and Elon Musk discuss and establish Starlink, and this has allowed us to connect across the country. And so we were able to take advantage of that in order to bridge that gap, to close that gap. That Ram flagged up as an issue. Also, women's empowerment and ensuring increased participation of women and all the more emphasis on rural women. Youth as well. We spoke with the doctor in a bilateral meeting how our government could use all necessary tools, expertise, everything available academically that UNESCO has in order to address these gaps, the gap that exists in some of our country's sectors. Although we have been making great efforts to fill those gaps, there is still— we still have a lot to learn and one of the fundamental principles that our country is working with is emphasis on international cooperation. And here we listen to UNESCO. We will take that to our president and we will see what is available to close those gaps that have been flagged up by UNESCO's RAM report. Thank you.

Thank you very much, Vice President. It's actually that last point which is very important. International cooperation, as you said, because the opportunities provided by AI that we all know, with the challenges that come with it, I think that it outpaces any institution, any single government. No one alone would be able to overcome these difficulties, these challenges of AI. I think that international cooperation is therefore essential. It's the missing link. And us, for us in the UN family, we have a major responsibility working between agencies And we have this beautiful testimony of it through the dialogue. We're going to continue in our UN family to do that through the specialized agencies, through the different country offices of the UN, in order to work closely with governments, with civil society, with the private sector, with scientists from around the world to make sure that we have AI which is for people, all people with no exception. I thank you very much. Vice President, Your Excellency, I thank you, ladies and gentlemen. I wish you an excellent working session.

Okay. Aha. That was fascinating. I want to thank very much our distinguished representatives for the fireside chat, His Excellency the Vice President and the Director-General. I thought that the insights around what is going on nationally in El Salvador was particularly valuable, and of course it is a pleasure of the UN to work with many partner countries around the world on these kinds of issues. I want to transition us now into the next couple of segments for the discussion this afternoon. It's going to be a series of 2 moderated panels, and we'll do one, and then we'll ask the co-chairs to come up and moderate a bit of discussion, and then we'll do the second panel, and then the same moderation. For the first panel, it's really a pleasure to invite my friend Shiko Gittal, Where did she go? Shiko Tagetao, who's the founder and CEO of Kala, to introduce and moderate the first panel. Shiko, over to you.

Thank you. Thank you. You can do from here or you can sit. Let me put this in here. Because I'm short, I will do that. Good afternoon. Aish. Good afternoon. Some energy. It's the first day of the conference. Good afternoon, everyone. Again, good afternoon. Good afternoon. Yes, some energy. It's my pleasure to be here. This is my first AI for Good, and of course, the first Global Dialogue. I'm really excited about this because we've been having this discussion for a long, long time, and it's time that the world comes together and speaks. For this session, I just want to give a bit of opening. Artificial intelligence is reshaping economies, public service, and societies, but its opportunities and risks are not shared equally. Nearly 3 quarters— put a mark on that— of the world is now online. Yes, even Africa. We are at 60%. Yet deep divides remain. Connectivity, compute, high-quality data, skills, financing, and institutional capacity. These gaps shape who gets to participate in and benefit from and help govern AI. And the prior side chart actually showed a bit of that. Our session tackles one of the most consequential questions in global AI governance: how do we ensure AI transforms potential rich and transformative potential reaches all countries and communities. And I'll double down on communities because Africa is a communal-led economy. Capacity building must move from beyond narrow. And I always tell people, if you've heard me speak, capacity building in AI should not just be just about how to use AI and ChatGPT. It needs to be tailored towards a coordinated approach towards building and creating AI for us by us. It means investments in skills, jobs, standards, open systems, and sustainable financing, where South-South collaboration and linguists, all our 2,000 languages in Africa and 7,000 languages worldwide, actually included. But beyond language, it's our culture and our context must be included, and the word is must. We have 25 minutes and we hope to be able to tackle some of this. I have an amazing panel. I was looking for them before this session started because I needed to see all their faces. I'll start with Pedro. Pedro is the Acting Secretary-General for UNCTAD. He leads the UN's voice on trade, technology, and the digital economy for developing countries. I did see Pedro. Okay. Yeah, he was. Next up is Gilmar. He's a member of the UN Independent Scientific Panel. He's also a researcher at one of my favorite labs, the Microsoft AI Research— AI for Good Lab in Nairobi, emphasis on Nairobi, researches on advancing locally relevant and data-driven solutions across Africa. The next few names I'm going to butcher because I'm from Kenya and my language is quite hard. Gilles Thonnet, he's the Deputy Secretary-General for International Electrotechnical Commissions. He shapes the global technical standards and underpinning trustworthy digital systems. Welcome. The next— this one is proper butchering— Arutyun Abestiani. He's a director of Ivanokov Institute for Systems. He'll tell us how it's said, his name and everything, and program from the Russian Academy of Sciences. They build open-source and full lifecycle infrastructure for trustworthy AI. I saw some of his work, quite impressive. Then there's Katharina Frey. She's the co-founder and executive director of ICANN. They lead a global initiative funding inclusive, sustainable AI for the Global South. And as I can see, there's something missing in my panel. There is no government representation. And I took matters on my hand and invited a government person. I want to invite Kanono Lamashonol. He's the Deputy Minister from Lesotho. But why he's in the panel— because there are many ministers from Africa on this session. Thank you so much for coming. He's the Chair of the AU Specialized Technical Committee on Communications and ICT. the experts panel. I've worked with him very closely on many issues, and he's very good at representing Africa. Welcome, everyone. And I will take the liberty of sitting because this is going to be a great conversation. So, As I mentioned, this is the scene-setting session. We want to be able to talk about capacity building beyond learning ChatGPT. Pedro, you start us off. We talk a lot about AI divide, yeah? From UNCTAD's perspective, what does that divide actually mean? Define for us divide, yeah? And is it widening or narrowing? And how do we close it is the most important question.

Thank you, Chico. Vice President, it's a pleasure to participate in this panel. Again, it's a privilege to address this Global Dialogue and let me go straight to the point. I will try to cover very briefly the debate from the investment perspective and from the cooperation perspective. On the investment perspective, I mean, to understand the AI divide, follow the money. Mm-hmm. We are living through what may become the biggest capital expenditure boom in modern history. Let me give you some figures on this. The world will spend close to $800 billion on AI infrastructure this year alone. This year alone. Cumulative spending will approach $8 trillion by the end of the decade. Memory chips are sold out through 2027. Companies are installing their own gas turbines at their data centers to power their own electricity. And strategic sectors like AI, semiconductors, critical minerals, clean energy now take half of all new investment in the world. Just to give you a reference, 5 years ago they took 20%. Now it's half of all new investments in the world. But the geography of this boom is, as you can imagine, and that's why we are talking about divide, is narrow. 2 economies host the majority of AI and semiconductor investments. Just 100 companies account for 40% of global corporate research and development. And of course, least developed, least developed and lower middle-income countries capture only 10% of strategic sector investments, half the share they capture in other industries. So as you can see, the barriers compound each other. A data center needs reliable power, water for cooling, high bandwidth connectivity, technical skills, and predictable data governance, all at once in the same place. Miss once, miss one of those, and the investment goes elsewhere else. So the— this investment divergence is upstream from all the AI divides, and in a CapEx boom this large, it can have huge effects down the line. This is some money I will maybe cover if we have time later, the the cooperation aspects, where it's not a question of money, it's a question of time.

Okay. I mean, from what I'm hearing from you, like, follow the money, there is no money in parts of the world to be able to invest in this AI, which is pretty sad. So, Kanono, you are the chair of the experts group for the AU. From a national and a regional perspective, what concerns led What countries let developing countries participate as AI consumers and co-creators? I'll need you to cover both. We are consuming right now, but we want to be co-creators. Every session I've gone today where there's an African minister speaking and a developing South-South minister speaking, they all want to be co-creators. But given what Pedro has painted, that looks like a very sad state of affair. How do we get there?

Thank you, Chukwu. Excellencies, colleagues, for us as a nation, we now understand that we have to meet some basic conditions like affordable connectivity. What we have learned in the past is that connectivity alone is not enough. In Lesotho, we have 100% connectivity, but only 50% of the people have access to internet. We also realized that reliable power is also very critical because some people have to walk distances just to charge their mobile phone. Of course, you talked about how do developing countries become consumers of AI and co-creators. One of the basic conditions is the digital skills, access to compute. What we have realized as the main barriers— of course, most of the people will say access to compute, but we have noticed that data is the main barrier. Mm-hmm. Our Deputy Prime Minister actually proposed that this forum should consider coming up with a framework to deal with unstructured and fragmented datasets, because most of the public sector data is either fragmented or unstructured. Those issues can be solved with— technically, they can be solved with implementation of interoperability platforms and the REC systems. But the question is, can they be trusted? But if we have a common framework on how to deal with such data, then every AI solution can find its way into every market. The second barrier is the skills. Unfortunately, AI does not depend on a single skill. Skill. You need specialists in cybersecurity, people who can manage data centers or even build data centers. If we can meet those 2 conditions, then we'll be able to drive demand. The demand will make compute available. What we have seen working in our environment is the investment in digital public infrastructure. and open source. Just last week, we participated in the Open Source Week in New York, where our developers interacted with their counterparts from across the world to learn and to network. Thank you.

Thank you. Thank you so much, Deputy Minister. I'm writing notes, and I'm seeing there's a story that is writing itself. Abebe, you're the researcher. You're everything that Kanono speaking about, right? You're the scientist, the researcher, you're building AI for Africa. How should linguists and— linguistics and cultural diversity be embedded in capacity building? And I know some of your work, so I will not put you on the spot, but I want you to speak from what's needed. Is it the datasets that he's mentioned about? Is it application? Is it local expertise, the skills part? Is it standards and frameworks that he's talked about? We need AI to serve our world, and I'm very specific to our world because it's serving other worlds. How do you make it happen? Yeah.

Thanks, Chico. As you've probably seen in our preliminary report, which is in your chair, we had one-liner that says, For AI to be successful, context is important. We said actually context decides outcome because we all are here with the agreement that AI is here, the technology is here. We could see what's possible. Ideally, we want to create an outcome which is positive, equally distributed across the region and populations. And for that to happen, of course, you need beyond the technology who access it, how we adopt it into existing workflows across institutions, how that diffuses into different sectors that would potentially make a positive impact. Beyond the technology to achieve that impact, what's very important is the context. Professor Wangari Maathai, which is a known Kenyan environmentalist and sustainability expert said, you may do the right thing, but without context, you are actually off track. I caught that because AI as a technology should refrain from repeating the same mistakes we had as many sustainability efforts in the past. That means we have 7,000 languages as discussed today, but it doesn't mean that the technology would equally serve all of them. That means we need to be more practically driven, context-dependent, so that the way we build purpose-driven, locally aware technology, that is a way to make an impact going forward. Nothing is more important to understand the context than what Shiko asked, culture and diversity. All these are the clear baseline to understand data, which is called knowledge, as the questioner mentioned. And for that to happen, let's not bring, you know, let's not assume an AI model that understands all the contexts at once for all languages. That will be very difficult. Rather, I'm an African. Given the different challenges we have, it's very important to understand the problem, the purpose, the actual resource, the infrastructure, and the outcome. That way of really bringing the promising technology into the actual problem we try to solve would be a best way forward. And as I said, context decides outcome, and we only understand the context when we bring different stakeholders from the government, from the private, from the youth, from grassroots communities, which are the driving forces in Africa, That way, we put people at the center of the design process, which enhances the chance of really bringing the technology for positive impact.

Awesome. Thank you. Thank you so much. That is amazing. Gils, we had a quick discussion about your questions, but what is the role of international standards in developing countries to become AI creators? I'm not giving you the the leeway to talk about consumerization, is how do we make— what are the standards that we need to put and frameworks we need to put in place for African and developing countries to be co-creators of AI?

Yeah, well, thank you for the question, Excellencies, dear colleagues. I would like to make the case for standards and international standards as one of the most practical tools actually for developing economies and developing countries to become, as you said, not only AI users but AI co-creators. But let me start by how we see the AI governance landscape. We see it in 5 normative layers, starting with the international law, international soft law, national law, international standards, and And then the normative power of what is possible technologically speaking, which is mostly shaped by the private sector. So the international standards actually form the link between those layers. They translate principles into practical tools. They provide the basis for regulation to governments, and they also give industry a common language to make products. Safer, to build interoperability, to make services and products more secure. So because the standards are developed by consensus, every country has an equal seat at the table regardless of their economic size and regardless of whether they have a big AI industry or not. So a country doesn't need actually a big AI footprint industry to shape the rules and the standards that will govern this area. So the participation itself builds the capacity, and they give national experts the knowledge they can use, they can leverage, they can adapt to build on these tools. If you look at the World Bank 2025 World Development Report, they call standards the invisible infrastructure. I think that's exactly right. The standards are largely hidden, they are unseen, but they power the world because they make, actually, products safer, they make products interoperable, and they make products operate across borders as well. But this invisible infrastructure relies on a very physical infrastructure. My neighbor mentioned data centers. for instance, energy-efficient data centers, the grid security, energy supply. All those elements are the base layers of AI. They're not compute, they're not, uh, you know, memory sticks. They are energy. Without reliable power, there's no AI. That's as simple as that. So that's why standards are really powerful bridge to co-create, and in particular also for developing countries, because the developers of AI, they don't start from scratch. They can leverage the collective experience we have in the standardization community. So at the IEC, we are committed to keep strengthening the participation of developing countries. We have training programs, we have dedicated actions. And actually, if we look at our AI standardization committee that we we operate jointly with ISO, more than a third of participating countries come from developing economies. So that's already a milestone. So I therefore encourage all the nations actually present here to participate, to connect with the international standards community, and to make your voice heard. That's really an investment in the invisible infrastructure that holds everything together, including AI.

Awesome. I like that. That's a beautiful case. So, Atun, as you tell us how to say your name and the name of your organization, is the idea— everybody's talking about data, energy, and things, but nobody has touched about trust. Yeah? And especially in a shared infrastructure, we talked about— somebody mentioned about DPI. So, my question to you is, we know no single country can build an entire trustworthy AI stack, secure, benchmarked, verified, all these things. How can open shared infrastructure, such as the international repository that you mentioned in your deck, help developing countries become genuine co-creators while each country develops common standards, building to— from what he said, through its own legal and cultural context?

Thank you for the question. I'm very glad to be here today, and really very interesting that several years scientists send message that trust is most important than each other question. But we need several years even in the AI area, not one month But maybe more than five years for achieving this kind of meeting. Thank you, thank you. It's very good. But let me speak one two minutes, but by Russian. Is it possible? Okay. Yeah. I will speak in Russian. The Russian language is one of the international cooperation languages, and I do cherish the hope that. In the future, AI will make it possible that we can each speak in our native language and that we understand each other without any issues. When I was a small child, for example, there was a fantasy film of the Soviet Union where that actually took place. People were standing next to each other and they were able to speak their native languages and everyone understood each other. There were no barriers. So today, I hope that even smaller languages will be covered by AI. Now, with regard to the subject and the question that was asked, well, In order to respond to any challenges, what's important here is to understand what problem we are facing. So over the last decade, what's been really actively developing is security lifecycle, as it's called. And we're of the— one of those who are developing that critically and trying to make sure our country is independent in that sphere. Over the last few years, we have seen that AI is presenting principally new challenges, which are not just data center issues, right? Not just— We need also new results analysis mechanisms, but also it impacts social development, deepfakes, of course, elections. And today, throughout the entire day, what we've heard is a number of examples of that and other impacts. Now, what I would like to say here is that for our country, we have a toolkit of technologies, and it's demonstrated that We can't just have individual campaigns, but we— or individual countries. We need to work together. So of course, the few years ahead will be an open range for us to work on this. Firstly, I think we need to fix at the UN level and perhaps at today's meetings level— and this is a great meeting for this— we have to understand if there are some sort of open models and open codes should be equally accessible to any student, any university throughout the world. The second point is that we need to have institutes which— and at a standardized level, we enshrine the fact that we work together on everything that links with security, because it's not just one tool, right? Today you can open any tool on the internet, but if you deploy it, you don't have enough security, you don't have enough knowledge to do that. It's a whole gamut of issues. And so we have to recognize this difficulty. One of the— one app is only one solution as part of this discussion. And it's just one that I'm mentioning here. We have many more. But is when dealing with AI, we're trying to— I'm sure you've seen in China, in the US, and also in Russia, We have to have labels, special labels, and we'll see that develop over the future because we have to have equitable access no matter the size of a company, the size of a country. That's the only way forward for humanity to develop because this is a global challenge. Once again, it's a global challenge, ladies and gentlemen. So it's a very important point. And another point I wanted to raise is that, and it's personally for me an important one, is that despite the fact that we're calling for singular standards for all nations to work together, and there's cultural specificities of each country, each country needs to solve those issues themselves because we can have a single standard, but everyone has their own culture, right? We need to take into account the cultural specificities, the cultural richness of each individual nation. There are different understandings here. For example, we heard from Saudi Arabia or Europe, Brazil, Russia, everybody has their own cultural specificities, right, their own culture. And we can't overlook this step in history. We have to have 100% reliable technology which would allow all at a national level, um, each individual country to solve the challenges that it's facing uniquely. Thank you very much.

Thank you so much for that. And to our last speaker, Katerina. I mean, we had a really interesting conversation with her, is how do we create sustainable and locally anchored private and public sector capacity building initiatives? And it's this locally anchored part that I'd love to hear from you. And then we need to talk, all of us, at some point, maybe not here, on how to finance everything we've spoken about.

Yes. Thank you so much. I feel I'm actually repeating what my neighbor has already said. I'm not so sure about you because it was so quick. I'm afraid I couldn't understand every, every single one of it. Now allow me to maybe close the panel as well by saying that I think AI, and it has been said before, is moving so quick, then I think we also need very agile networks. So at ICANN, so at the International Competition and AI Network, what we have tried since 2 and a half years now is really to connect the different actors because there's so much already out there. Many of it has been mentioned by my previous speakers, but what you basically need is you need access to a large, very powerful compute. You need access to knowledge or qualitative datasets. And very importantly, you also need access to experts, both domain but also AI experts. And Our members are all academic partners. I think that's also important. Sometimes when you have global divides, it's helpful to work amongst researchers. We do have member states like Switzerland and also the Finnish Foreign Ministry have been patrons— I don't know how you say that in English— of ICANN. But it's basically we connect demand and offer. So give me— allow me to give me 2 or 3 examples One is our founding members of Data Science Africa. You have mentioned in the African continent, grassroots networks are very important. Deep learning in DARPA, we're discussing as well. So they have led work on AI weather forecasting systems for the farming communities. In this case specifically, they didn't have access to sufficient amount of compute. We have linked them to the Swiss supercomputing centers. And then they created a fantastic research collaboration that then leads into direct impact locally. If we move around from, from countries, I think we are here at the UN. Also, the UN organizations have needs. Sometimes they want to have their own independent solution, maybe not being dependent on one specific hyperscaler. So we have also, with the support of the Swiss Foreign Ministry, we had just launched 2 months ago a program for the UN organizations, specifically the ones in Geneva, and we were completely overrun. So basically what we did, we linked different UN organizations, I think yours as well, to researchers from the whole ICANN network, which is basically from, from the African continent, from, from the European continent, and in Asia we have one member so far, the NTU Singapore. And they— you, you— so you see there's a huge demand. of co-creating your own AI solutions. I could continue. I think we have to end, but I think if you allow me to also reach out to the room, you've seen Latin America is not being represented yet, so we would be more than happy to have also academic partners from Latin America, Middle East, and other parts of the world. Thank you so much.

Thank you so much, everyone. So, to just give a quick summary, we said follow the money. Connectivity, power, data, and skills are important. Context— I love that sentence— context decides outcomes. Purpose-driven, locally aware AI is what is important. The case for standardization: participation creates capacity. I loved that line. And I got the last part of the Russian translation, and it said, We must push for equitable access. Trustworthy takes time. Trust takes time. And that's something I think the AI world needs to understand is trust creation takes time. And then you finished with, we have to start with the grassroots. Thank you very much, everyone. And thank you very much, my panel.

Thank you very much, Shiko, for running a very interesting panel. Can we just appreciate them again, please? Thank you. Great. Now, my pleasure. I think we're starting to get into this issue of AI divides. I mentioned at the beginning there are many AI divides, and we're starting to hear some of those come out of the panel. And I expect that we'll continue to hear about those as we move into this next session, or next part of the session, which is the moderated discussion. And it's my pleasure to invite again our co-chairs to come sit at the podium at the front. His Excellency Minister Diouf and Mr. Jovan Kurbalija. Thank you. Hello everyone, and thank you for your patience. We will now facilitate interventions for this segment. Statements will be delivered based on the speakers list established through inscription that was made available on the UN Global Dialogue on AI website. Please note that in the interest of promoting broad participation from stakeholders, we will proceed with one representative from a member state and one representative from other stakeholders, and we will alternate back and forth. All speakers will be invited to intervene from the floor where you are seated when you are called upon. Please use the microphones that are available at the desk. Alternatively, if you are not seated at a microphone, you may make use of the center stand mic that is in the aisle here as you enter the room. The e-delegate list is now closed, and any changes to that list must be communicated to the secretariat in the room that's next to the podium. Very importantly, a timer has been set and is visible on the screens here. We ask that all speakers respect the allotted time so that the largest possible number of speakers may be accommodated during this segment. Thank you very much, and I now turn it over to our distinguished co-chairs.

Thank you so much. So let's move to our high-level government speakers, and I would invite our representative from South Africa, Deputy Minister of Communication of Digital Technologies.

Thank you. Thank you, Chairperson, Excellencies, distinguished guests, delegates, ladies and gentlemen. South Africa welcomes the opportunity to contribute to this important discussion. Artificial intelligence is arguably the defining technology of our time. However, the benefits of AI will not be shared equitably unless we address the deep inequalities that continue characterize the global digital landscape. As South Africa, we believe that AI must be a force for inclusive development, economic transformation, and social progress. To achieve this, we must first close the gaps that prevent many countries and communities from participating meaningfully in the AI ecosystem. The first priority is digital infrastructure. AI cannot flourish where there is limited connectivity, insufficient broadband access, unreliable electricity, or inadequate computing resources. Investment in universal connectivity, data infrastructure, cloud computing capacity, and digital public infrastructure must therefore be viewed as essential enablers of AI adoption. The second is priority— the second priority is human capacity development. The success of AI will ultimately depend on people, and we must invest in skills as developing countries require not only AI users but also AI developers, researchers, entrepreneurs, policymakers capable of shaping the future of this technology. South Africa's approach places significant emphasis on strengthening AI skills across government, industry, academia, and society. The third priority is access to data, computing resources, and innovation opportunities. Today, many developing nations face constraints in accessing datasets, computational infrastructure, and financing needed to build local AI solutions. If left unaddressed, these inequalities risk concentrating AI innovation in a small number of countries and corporations. We therefore support greater international cooperation, partnership, knowledge sharing, and investment mechanisms that enable broader participation in the AI economy. Equally important is ensuring that AI reflects the diversity of the world's people. Building local datasets and supporting indigenous innovations are critical to ensuring that AI solutions are relevant, effective, and inclusive. Bridging the AI divide is not only a matter of technology, it is a matter of fairness, opportunity, and sustainable development. Our country believes that no country should be left behind in the AI era, We therefore call for strengthened global cooperation. Thank you so much.

We just have 3 minutes. Okay. You mentioned about investment. You mentioned about investment, human capacity development, access to data. All this, we are aligned on this vision, and for sure we'll take care about it. The next stakeholder is from Interpol, Mr. Huanzhang Fu.

Okay, good afternoon, Excellencies. Can you raise your hand? I'm here. Okay, thank you. Sorry. Excellencies, distinguished delegates, ladies and gentlemen, it is my great honor to be part of this dialogue. So, you know, AI is rapidly transforming every aspect of the society, including the way law enforcement protects communities and combats increasingly sophisticated crime. Yet, as AI capabilities advance, AI literacy and readiness among law enforcement agencies vary considerably across countries. Bridging the AI divide is therefore not only a technological challenge but also a capacity-building imperative. So at Interpol, we believe that effective AI adoption begins with knowledge, trust, and responsible governance. Technology alone is not enough. Law enforcement agencies should have the skills to understand AI's opportunities and limitations, the governance frameworks to deploy it responsibly, and the confidence to use it in ways that uphold the rule of law and public trust. These are the foundations for meaningful and sustainable AI adoption. Together with UN partners UNICRIR, Interpol has been actively supporting this objective. In 2024, we jointly published the Revised Toolkit for Responsible AI Innovation in Law Enforcement, providing practical guidance for police agencies around the world to adopt AI in a lawful and responsible manner. Building on this foundation, Interpol and Unicry are currently implementing the EU-funded AIR project, which seeks to further embed responsible AI innovation across the global law enforcement community. One of its key outcomes is the development of practical training programs that strengthen institutional and human capacities. To support this effort, we launched the TRAIL e-learning program, Training on Responsible AI Innovation in Law Enforcement. Which is freely accessible through Interpol's learning platform for law enforcement in all member countries. We also delivered webinars and in-person training programs to officers from different regions, helping translate principles into operational practice, including CARICOM, CEPOL, OSCE, just to name a few. Looking ahead, we remain committed to working closely with the United Nations member countries, academia, and industry to strengthen global AI capacities for law enforcement. By sharing knowledge, developing practical guidance, and fostering international cooperation, we can help ensure that AI becomes a force for safer communities and stronger international cooperation. Because bridging the AI divide is not simply about expanding access to technology, it is about empowering people and institutions to use AI responsibly, effectively, and with trust. Thank you.

Thank you. Thank you for bringing us institutional capacity building as important. Also, safety and trust is a critical aspect. Our next speaker is Her Excellency Niamh Smyth, Minister of State at the Department of Enterprise Ireland. Please raise your hand and Please go ahead.

Excellencies, distinguished co-chairs, distinguished guests, ladies and gentlemen, AI presents a significant opportunity to drive human development and economic growth, enabling individuals and communities to fully participate in economic, social, cultural, and civic life in the ways encapsulated in the SDGs. But we see the benefits of digital technologies are often not experienced equitably, either within societies or between nations globally. If left unregulated, AI risks exacerbating these divides. Governments, businesses of all sizes, workers, civil society, and communities must all be empowered to understand and consider how AI will affect us so that together we can shape its trajectory. In Ireland, we are working to ensure that all citizens communities and small businesses have the digital skills needed to thrive, as well as equitable access to technology, essential public services and high-speed broadband. Our National Digital and AI Strategy targets socio-economic, enterprise and geographic divides through infrastructure investments, business supports, digital literacy and other initiatives. It aims to ensure that SMEs remain competitive to empower citizens to access and to engage with technology developments, and to guarantee equitable access to essential public services. Internationally, we must recognise the potential benefits of AI are not universally available, and countries with less developed AI infrastructure often have less influence over how AI is governed, even when the impacts on their populations are significant. Open-sourced and open-weight AI models can play an important role in widening access to AI capabilities, supporting local innovation, language diversity, digital sovereignty. And access to open AI models expands— governing frameworks must also evolve. Ireland has a focus on inclusive multi-stakeholder models of governance which are risk-based and take a human-centric and human rights-based approach. No single entity, whether government, tech company, academia, or civil society, can alone navigate the possible— the complex societal and ethical impacts of AI. The UN can provide a critical platform for the digital cooperation, the global digital cooperation we need to address these urgent questions. All voices need to be included in the conversation on global AI governance. to ensure that technology is safe, equitable, accessible, and inclusive. The Pact for the Future affirms that bridging digital divides is essential in achieving SDGs thriving progress and growth and allowing individuals and communities to fully participate in society. Within these— within digital divides must be addressed urgently through meaningful global participation, and I hope— Thank you.

Thank you, Minister. The technology doesn't have any mercy, but it was so great— it's so great to hear from you. Ireland is an important player and hub of data and digital developments, and for bringing, for example, issue of open source on open weights, which is one of the topics that will be Our next speaker is coming from Banjul, my former students from Joko Lab's CEO, Ponceleite Ilalei. Ponceleite, where are you? No? Is Ponceleite in the room? No? It's a pity. He's my former student. I expected him to be here. Okay, then the next speaker The next speaker is from Golden Gate University, Smriti Goudhaman, Global Professor of Practice. Just raise your hand if you're in the room. Yes. He's the guy. No, no, he's a government representative. No, it seems that other stakeholders apart from governments are not present. Then we have the stakeholder from Bangladesh, Ahm Bazlur Rahman. Chief Executive Officer, please go ahead.

Thank you, sir. Moderator, distinguished participants, and Excellencies, let me add one simple question. What happens when AI reaches a community before skills, trusted information, safeguards, and institutional capacity reach the community? The digital divide does not disappear. It becomes an artificial intelligence divide. I come from Bangladesh, where millions of people live in rural, coastal, and climate-vulnerable communities. For them, the AI divide is not simply connectivity or computing power. It's a divide in language, data, skills, reliable information, institutional capacity, and power. The problem is too much AI capacity building still treats developing countries and communities mainly as user and consumer of technology, but our ambition must be greater. This is the real AI capacity gap. It is not only technology gap, it is an institutional gap, it's a language gap, it's a financial gap, and it's ultimately a power gap in line with the political gap. So I propose 3 concrete priorities for the, for the international community. First, build locally anchored AI capacity ecosystem. Second, invest in linguistic and cultural inclusion. Third, create sustainable and accessible financing. The fund should support local language AI literacy, synthetic content verification, prevention of technology— technology-facilitated gender-based violence, public interest AI, and meaningful participation in AI governance. My message is very clear: access without capacity creates dependency, capacity without local ownership does not survive project cycle, and AI governance without linguistic culture and community participation will reproduce a very divide we are here to bridge. AI should not simply reach communities.

Thank you, sir. Thank you so much. Our next representative is Her Excellency— His Excellency from Japan, Yohi Onishi, Yohi Onishi

from Japan. Excellencies, it is my great honor to have this opportunity to deliver a statement on behalf of the Government of Japan. AI has significant potential to contribute to the achievement of Sustainable Development Goals, including by promoting economic growth, enhancing productivity, improving education and healthcare services, and addressing disaster risk reduction and climate change. However, the benefits of AI are not enjoyed equally around the world. Many countries face challenges in terms of human resource capacity, computing resources, and data. Institutional development and digital infrastructure. We must not allow disparities in AI utilization capabilities to give rise to a new digital divide. Japan is co-creating a safe, secure, and trustworthy AI ecosystem with the Global South through international cooperation based on trust. Cooperation toward the cultivation of AI talent and capacity building in particular provide the foundation underpinning such an ecosystem. For example, Japan is supporting the capacity development of a wide range of human resources in Global South countries such as ASEAN and Africa, including through the use of various cooperation tools. We are providing such support not only to AI engineers but also to policymakers, government officials, and education professionals. A defining feature of Japan's international cooperation is that it helps building foundations that enable each country to harness AI based on its own ownership and development strategies. This approach promotes close Collaboration on joint research, the development of AI that reflects each country's languages and cultures, and the co-creation of AI solutions to address the social challenges that each country faces. Achieving concrete results requires the engagement of multi-stakeholders, including the private sector and academia. Japan welcomes the participation of a diverse range of stakeholders in this global dialogue. Japan will continue to deepen cooperation with all stakeholders and work on co-creating AI ecosystems. tailored to the realities of each country and region. Thank

you so much. Our next stakeholder is Shikodayi— Shikot.ai. Christiana. Is Christiana around? No? The next one is from Moscow, Anna Abramova. Anna is not around. The next one is from AWS, Joshua. Mr. Joshua? Yes. Not here? So the next one is from DPGA, Leah Gimpel. Leah is not around. So where are you guys? The next one is from NetHope, Misha Parker. Michelle is not around as well. So the other one is from UN Permanent Forum, Walt Ernest

Trist. I am. Distinguished colleagues, As noted in the preliminary report presented earlier today, most of the world's languages and cultures remain underserved, with AI development concentrated among a handful of the world's major languages. The report also finds that of the 7,000 languages, just 1,000 have the foundation needed for meaningful inclusion in AI systems. To put it differently, 6,000 language communities the absolute majority of the world's linguistic and cultural diversity lack those foundations. These are communities that already face severe challenges, the majority of them indigenous or speaking languages that are severely or critically endangered. The report points to another more alarming aspect. While AI develops at unprecedented speed, the performance gap between dominant and underrepresented languages persists and is not narrowing. This means there is a critical barrier that those 6,000 language communities cannot overcome alone. In addition to critical lack of resources and attention, one critical obstacle stands out. Developing AI for such communities is methodologically distinct. The common assumption that AI can be developed for large user groups and then scaled down does not hold. Because the building blocks, data ecosystems, and usage patterns of these communities differ from mainstream ones. Despite their number, a critical lack of knowledge remains about the methodology of building AI for such languages and cultures. This is a systemic issue that goes beyond AI and technology alone. It is a question of access, not merely to AI, but to the digital world itself and to responsible and responsive AI that supports these communities rather than marginalizing them further. 3 parts could help close this gap. First, communities at the table. Indigenous peoples and other under-resourced language communities must be present when AI's development is discussed and shaped. Researchers, developers, and policymakers need to listen and hear these communities, both to understand their realities feedback and to strengthen AI's own capabilities. Second, dedicated research, development, and resourcing internationally and nationally, including technologically advanced countries, to explore how AI can be developed for underpowered communities. Closing this gap helps to build a more responsible and responsive AI ecosystems as a whole. Third, governance of data and technology. Not simply because these communities are right holders entitled to stewardship of their data, knowledge, and ways of life, but because such governance allows AI to be fit for purpose, non-extractive, and supportive of communities struggling to compete with the world's major actors.

Thank you. Thank you for reminding us of the relevance of Indigenous communities and knowledge which is probably not codified in texts, books, big writings, but equally important, especially wisdom of those communities. And I think it's a reminder for all of us that this knowledge is critical for not only for them but also for humanity as a whole. Our next speaker is Ambassador to the United Nations of Moldova, His Excellency Vladimir Tzuc. Please, Ambassador. Yes, I'm here. You're here. Please

go ahead. Thank you. Thank you very much. Well, first of all, I believe this discussion indeed reflects what the independent scientific panel has described in its report. It's about gaps, and gaps are creating disparities. It's governance gaps, it's resource gaps, capacity gaps. They create disparities and unequal access. It's also implication for human rights, for security, for democracy, for environment, increasingly. And no country can deal with it on its own, but also, and we should also not allow that only some limited group of actors are controlling the AI. I think these are the general principles which are pretty much supported by everyone in the room. So we believe this dialogue is very timely, and we hope this will go and lead to a concrete outcome. My country, Republic of Moldova, is the advocate and the supporter of global governance of AI. We've shown it in the Council of Europe when we were among the proposers of the first framework convention on democracy, human rights, rule of law, and AI, and we were among the earliest signatures of this. We also adhere to the UNESCO recommendation on ethics of AI, and we will be promoting human rights-based approach to AI during our Human Rights Council membership, but also in our presence, current presence in UNESCO Executive Board. We're also EU candidate countries, and we are working now on adopting our legislation to EU, so transposing EU AI Act. We will create an agency. We will follow up with the program. We have undergone the UNESCO Readiness Assessment Report. Report. And what it discovered, that yes, we are— we have an ecosystem, it's developing, but we lack the capacity. It's in public sector, in private sector, in academia. So the lack of capacity is an issue, not for some areas, not only for Global South, also for Europe, for many countries. It's an issue which we would like to compensate. So that is why we believe that capacity building, investing should be at the core of AI dialogue and AI cooperation, international cooperation. So this is how it will make AI less disruptive and more transformative. This is what we should aim at. We welcome the announcement that the next AI dialogue will be in May next year in New York. So we hope Really, the discussion there will become more practical, focused on initiatives to bridge the capacity gap. And I believe one of the takeaways which you can— we can suggest for your outcome for this discussion, invest in public-private partnerships. The private companies, they own mostly now AI. Government are regulating. We need the partnership and we need everyone on board and all the stakeholders as well on board and at the table. voicing their concerns and proposals. Thank you very

much. Thank you, Ambassador, for this timely reminder that capacity is needed everywhere, not only in developing countries but in many developed countries, in institutions, and that this concern is shared across societies worldwide. Our next speaker is ICAM— CEO of ICAM, Talangana AI Innovation Hub. Raise your hand. No? We have to see what we are going to do with stakeholder participation. They're missing. Then next speaker is from ASEAN Foundation, Diera Paksi, project manager. Please go ahead. Yeah. Oh, sorry.

Good afternoon, Excellencies, distinguished guests. My name is Dira. I'm currently serving as a project manager for AI Ready ASEAN program, one of the largest AI initiatives across ASEAN. To date, we aim to empower around 5.5 million individuals comprised of youth, parents, and also educators through 6 key initiatives such as awareness raising on AI, development of platform, training of trainers, policy roundtable discussion, And also development of the research and outlook. So I think my point is, today we've seen that AI capacity building is not simply about increasing AI literacy. It's about building an ecosystem that enables government, educators, industry, academia, and communities. So from ASEAN's perspective, 4 conditions are essential. First, human readiness. Investment should not only focus on the infrastructure, but also on educators, students, policymakers, and also communities. Second, local partnership. Local organizations understand cultural context better than regional organizations. That's why we are working with the 23 civil society organizations across ASEAN. And third, government ownership, so capacity building becomes sustainable only when governments are involved from the beginning. We have a policy roundtable discussion to combine the policymakers together with our local implementing partners to develop policy to ensure the sustainability of the program. As of now, to date, we have empowered 8 million individuals across ASEAN. The numbers continue to grow through multi-stakeholder partnership. I think, based on our recommendation, We recognize human readiness as a core pillar of AI governance alongside innovation, safety, and regulations, while we need to also promote regional multi-stakeholder partnership that combine government, academia, industry, civil society to deliver locally relevant AI capacity building. This is the one that we have built so far. And lastly, invest in sustainable education ecosystem that strengthen not only not only the people, but also specifically to the teachers, students, institutions, policymakers, rather than focusing on the short-term AI training. So thank you so much, Excellency. So yeah.

Thank you so much. So we're gonna have another 2 speakers before moving to the second panel. So from Her Excellency from Bangladesh, Nahida.

Thank you, Chair. Excellencies, distinguished delegates, bridging the AI divide is ultimately about bridging the development divide. If AI is to become a force for sustainable development, it must be accessible, inclusive, and governed in a manner that upholds human rights and places people at the center. The AI revolution should accelerate development, not deepen existing inequalities. It must empower all people regardless of nationality, ethnicity, gender, age, or income. AI must benefit everyone through education, employment, entrepreneurship, and innovation. Bangladesh is investing in digital foundations to accelerate inclusive development. Government is taking initiatives to make mobile internet one of the most affordable in the world. Through more than 239 million mobile financial service accounts, we are expanding financial inclusion. Our digital public infrastructure is enabling innovation in health, agriculture, education, disaster response, and social protection. Yet infrastructure alone is not the answer. Frontier AI requires enormous computing power and investment, while AI talent remains concentrated in a handful of technology hubs. Most AI models are trained on data, languages, and context from the Global North, often producing biased outcomes. Distinguished delegates, allow me to propose a few actionable paths. Pathways. First, democratize access to AI. We should operationalize the proposed Global Fund on AI and establish concessional pathways for LDCs to access AI compute and frontier technologies. Second, invest in building lightweight, affordable, resource-efficient, open-source AI models trained in diverse languages data, culture, and contexts of Global South to democratize technology and prevent digital exclusion. Third, invest in people and local innovation. We should establish local AI innovation hubs, strengthen AI skills, support startups in developing countries. Fourth, ensure meaningful representation of LDCs in global AI governance. And standard-setting bodies. LDCs must become co-authors of global AI governance rather than recipients of decisions made elsewhere. Finally, for Bangladesh and many developing countries, implementing the aspirations of the Global Digital Compact is a key development imperative. With a pro-development and inclusive governance ecosystem, AI can augment human capabilities create decent work, and raise productivity rather than replacing human judgment. Thank you.

Thank you. Thank you so much. I really like this sentence when you're saying we must— AI must be benefit for everyone. I really like this sentence. We cannot have AI without internet access. So our government should invest and ask people, especially for operators, to invest on the access, the universal access. Our next stakeholder is Xinhua Lao School. Wei Xing Sheng? He's not here. The next one is the Smart City Smart City, Crystal Ann. Crystal is here. Yes, thank you, co-chairs. Thank you, Your Excellencies.

Most sessions frame AI and digital divides as a pipeline problem— more skilling, more training, and more capacity building. But we believe that that particular framing is incomplete. We at Smart City have trained over 400,000 Filipinos across 60+ local government units in the Philippines, and I would say that the pipeline is rarely empty. What the pipeline thinking misses is what happens after the training. So what are the 3 gaps the pipeline training doesn't capture? Number one, absorption. We train people, but labor markets and procurement systems have no slot for them. Local government graduates AI-literate staff with no budget to hire or deploy them. Training without a destination is a holding pattern, not capacity building. Number 2, a seat at the table. Communities that build AI literacy at scale are still absent from the rooms where standards, data sharing rules, and governance frameworks get written, which shows that using AI is not the same as shaping it. 3, the underlying assets: compute, data, and the institutional way to negotiate for both. No volume of training changes who owns the infrastructure AI runs on. With this, we believe that bridging AI divides, or digital divide per se, is not a training problem with a training solution. It is a power and access problem. Calling it a pipeline issue lets us— let us declare success once training numbers look good, while the structural question stays unanswered. For the co-chair summary and for this room summary, 3 priorities I would like to put forward. Number one, tie capacity building to financing and to absorption. as well— jobs, procurement, enterprise pathways, and not only completions. Guarantee implementing communities a formal seat in AI standard setting. And number 3, treat compute and data access as core to capacity building and not just adjacent to it. Thank you very much.

Thank you very much. So we're gonna let our Panelists, the second, the second panelists, okay, to come here. So after that, we'll come for this last round. Thank you.

This round of excellent comments from the floor, interventions from the floor, and thank you to our co-chairs, His Excellency Mr. Diop and Mr. Jovan Kurbelija. It is now my pleasure to kick off today's second panel by inviting the moderator, Ms. Crystal Rugegi, who's the Managing Director of C4IR Rwanda, to join me on stage. Thank you. And Ms. Ruggieri will introduce the members of the next panel. Thank you very much.

All right. Good afternoon, everyone. I know we're running a little bit behind schedule, so thank you for your patience. As she mentioned, my name is Crystal Crystal Rugege. I'm the managing director of the Rwanda Center for the Fourth Industrial Revolution, and it's my honor to serve as the moderator today. So, as we heard earlier, you know, the first panel was really touching on mapping the barriers and really seeing what does it take to bridge the AI divides. And I think this afternoon in our panel, what we want to do is really touch on the solutions. What will it take to actually close those gaps? So that's what we'll be doing over the next 25 minutes. If you'll allow me to introduce my esteemed panelists, I hope they're all in the room. I'm looking for some of them. But starting with His Excellency Sid Ali Zouraki, Minister of Post and Telecommunications of Algeria. You're most welcome. Ambassador Eugenio Vargas, Garcia, Brazil's Ambassador for Technology and Innovation. You're most welcome, Ambassador. Thank you. Professor Vukosi Marivate. He's one of the members of the independent international scientific panel on AI that we heard from this morning. Alessandra Sala, Senior Director at Shutterstock and Global President of Women in AI. Thank you. You're most welcome. And lastly, but not least, Urvashi Anija, founder and director of Digital Futures Lab. Please join me in welcoming them. Yes. Okay. So welcome again to our panelists. So, Your Excellency, I will start with you. You know, Algeria has made digital foundations a national priority, and we really commend you on that. You know, from connectivity to data infrastructure and skills. And so, from a government perspective, what conditions are necessary for developing countries to have equitable choices and opportunities, not only to use AI, but also to develop, adapt, evaluate, govern, and shape it as responsible co-creators?

Thank you. So for Algeria, the first point is very clear, actually. There can be no meaningful AI governance without strong digital foundation. In fact, if a country does not have connectivity, does not have trusted data governance, cybersecurity, digital skills, cloud capacity or capability, research, and development, access to computing capacity. It will be asked to govern a technology that it cannot be fully shaped by this same country. So definitely it's not an inclusion, it's dependency. So here for developing countries, the challenge is not to use AI application. The challenge, or the real challenge, is how to build sovereignty capacity. We can see today how AI is guiding us. We are following it. So most of the country— and today there's only 2 countries that are building AI or shaping the AI, while all other countries are just following the trend, with my respect to each one, everyone. So adapting national priorities is a must. Evaluating the risk and protect citizen is our priority also. Practice and having standards and governance framework that will shape the future, future of the future generation. Thank you.

Thank you. Thank you, Your Excellency. I'd like to turn next to Urvashi from the Digital Futures Lab. You know, at your lab, you've studied how AI actually lands in the societies of the global majority and often beyond the headline numbers. And so, you know, we often frame the AI divide in terms of compute, connectivity, and skills. And from your work, is there a divide that we're not paying enough attention to? If you can shed light

on that. Thank you, and thank you for inviting me to be on this panel. I think the gap that we don't often talk about, but it's been very reassuring to hear it actually talked about a number of times today is the oversight gap. And that gap is increasing as AI systems are becoming more complex. And this space of oversight is also something that is being routinely underinvested in the Global South by governments, by funders, by development agencies, even as we push ahead with integrating AI into critical social sectors. And this gap, this oversight gap, I think runs at various levels. At the level of the state, evaluation, procurement standards, incident reporting, continuous monitoring. All of these remain underinvested areas. The Global South is seen as an area— is seen where we can use AI to leapfrog complex social economic development problems. But in the process, we're also skipping over the crucial step needed to build the relevant institutions that can enable that oversight for all. And building that domestic evaluation capacity and infrastructure is a way that countries in the Global South can ensure that AI systems actually promote local values and priorities. It's a key part of the AI stack, and it's a key way in which states can actually extend sovereignty, but that we don't talk about when we're talking about the digital divide as well as talking about AI sovereignty. And then I think that that oversight gap extends to the societal level as well. Civil society and independent media, these are the checks and balances that make oversight real. And so where civil society can't flourish, we end up borrowing frameworks from the Global North instead of building our own. And without independent resource media, there's no one building the public literacy that accountability depends on. And I want to just end by saying that these divides on oversight actually compound the innovation divide. Because if we don't build that oversight mechanisms, if we don't build that evaluation infrastructure, we're actually harming people on the ground. And we're reducing public trust in AI systems. And slowing down adoption. So the safety gap or the oversight gap is widening the innovation gap. It's— and oversight is not only about risk management, but it's also about the foundations for inclusive development. Thank

you. Thank you. Thank you, Avashi. Professor Bukosi, my friend, you've actually been doing a lot of the work that's that she's been talking about, not just recently with the AI scientific panel, but I think a lot of the work that you've been doing for many years on the continent of Africa. And you've also been, you know, you spent a number of years building open community-driven AI research for African languages in particular. And, you know, as we look across, you know, the African continent, we see that, much of the heaviest lifting in AI capacity building has been done by grassroots organizations like yours. And so, how can we better connect global majority countries with their local ecosystems to give those communities doing the heavy lifting a real seat at the table by bringing them

together? Yeah, thanks, Crystal. And also, I'm very honored to be on this panel. So, in some ways, we can go back about 11 years in 2015, and on the African continent, there was brewing of a lot of people thinking about, hey, we see these kind of ongoing AI revolutions, and we also see what's also coming on the horizon. What are we doing? Are we getting ready? On one part is that, you could sit and say, we're not building and we're not going to build, but on the other is a number of young people started then saying, how do we then build ourselves, do the capacity building, build up capability? And that's where, on one side, in Nyeri in Kenya, you've got Data Science Africa starting in 2016, the beginnings of the ideas of the deep learning in Daba. Now the— I just

want to— If I can get you to hold your thought for one second. If you can help me in joining us to welcome His Excellency the UN Secretary-General for honoring us with his presence for this panel. You're most welcome. Yes, so if I can invite you to complete your thought.

Yeah, yeah. Thank you. Yes, so seeing like the Deep Learning in Daba now, which has become the largest gathering of research, innovation, and development on AI and data science across the continent starting in 2017. The thing I would suggest is on 2 sides. On one, for policymakers to engage with these communities. There's many of them, but they do work in tandem to try and create a space where if you have young people— remembering that Africa, half the population is 19 or younger, right? They're looking for what are going to be the skills of the future. So connecting with these grassroots organizations allows access of understanding the needs and then also looking at the future. For the researchers and developers, academics, On the other side is you also must be involved in the policymaking. I applied to be part of this AI panel because of that, of knowing that, yes, I might be one of the co-founders of the Deep Learning Endowment, but I left it. I left as a board member at the end of last year, hoping that then I would be accepted to be on the panel. To take up my next responsibility for the continent. But these are things that people should be doing so that we can reduce that gap in understanding and enable more and more of using what we have. To the point I wanted to say to the minister, we have an IndabaX Algeria, and I am hoping after this you will connect to that community that is based in Algeria, really doing so that you can see people are building in Algeria as well. well.

Yeah. Thanks, Fokosi. I'd like to turn next to Alessandra. You champion the human side of these divides, and that means the skills and the workforce that's really behind this technology. So, how can capacity-building efforts in AI skills and AI literacy equip individuals, institutions, and workforces to adapt to and shape the future of employment in an AI-driven economy.

Thank you, Crystal. It's an honor to be here with these panelists and with esteemed audience that we have today. I'd like to start offering a practical perspective. At Shutterstock, I'm leading the AI transformation, and one of the pillars is literacy among 8 others. At Women in AI, as president of Women in AI, we devote all our volunteering work to bring more women into skills, AI core skills, so that we can be at the table with the right background and with the right capabilities. And at Shutterstock, one thing that while developing this literacy program I have experienced during my office hour. They come to me and they say, Ale, can you help me to use Copilot and to save time, or any other AI agent? And I turn at them and say, do you want me to fire you too? And they look at me like this, say, no, today we'll do something different. We'll use the right tool that you need to solve your problems and will reinvent your workflow. So just to give you a very practical example, a person in sales comes to me, wanted to save time. I say, no, we will reach your customer with personalized offer. I will help you to reach out instead of 50 customers a day, 500 customers customer a day. So we go to scale together, and I'm enabling you to analyze your customer base so that you can reach those that are not writing to you, but those are in need of your support. So we move you from proactive work to— from reactive work to proactive work. So at the end of the session, we haven't saved anytime. We have just moved from a different way of working and doing your job. Why am I sharing this story? Because we should, as leaders in this room and in government, think that AI can be used to replace employees and save time and just follow these stories and these narratives that we have today, or think critically and design the workforce that we want to achieve. Each of us as an employer has a purpose, and we can augment, empower, and strengthen that purpose, and therefore creative society. society, creating a workforce that works for all. Thank you.

Thank you, Alessandra. Ambassador, if I can turn to you. Brazil has championed AI cooperation through several multilateral fora, and from that experience, what does it take for international cooperation to give developing countries equitable choices? and opportunities to use and adapt AI and shape it in their context, so they're not just consumers but also co-creators.

Thank you. Thank you for the invitation. I'm glad to be here. Yes, you mentioned equitable choices, because I think we could perhaps think of digital sovereignty as something important to allow developing countries in particular to make sovereign choices about how they can invest, adopt, or have a more proactive role in AI development. Because I think it was said before that mastering the full AI stack is not for everyone. Maybe 2 countries in the world would perhaps be in a position to do So if this means complete self-sufficiency, and when we talk about digital sovereignty, it's not to be regarded as isolationism, that we are closing our borders or seeking to produce everything locally, but it's rather a continuous process of reducing dependencies. Some of them are structural dependencies. And we should gradually expand our political agency within the global digital ecosystem. This is a progressive increase in technological autonomy in a selective and incremental manner, because these countries should primarily focus on critical layers of infrastructure, and by doing so, strengthening their national capabilities through international cooperation. That's key for your question. That's the point. International cooperation is like a strategic shortcut to access global knowledge, infrastructure, innovation networks. This is key for digital sovereignty as well. Because these countries will seek diversification of partners and suppliers. They could develop national alternatives if this is possible, fostering domestic solutions for strategic products or digital services, particularly in technological niches where they could have some comparative advantages. And some call this de-risking as well, because we're trying to reduce vulnerabilities in these countries, implementing measures to enhance their digital resilience. To conclude, I think also technology transfer is still important, but under favorable conditions, and developing countries should seek international partners and talking to them, negotiating on a case-by-case basis so that they can have concrete trade-offs and then a win-win cooperation. And finally, developing countries should be more active in terms of their participation in international discussions in this regard. And this is— it means elevating the voice of the Global South to reduce asymmetries and democratize access to technology. I'll stop here. Thank you.

Thank you, Ambassador. So, as I mentioned, we're running a little bit short on time, so I want to use the last few minutes to give our co-chairs some concrete concrete material. You know, they have to take this cluster and really distill all of the wonderful insights into 2 or 3 top priority actions. And so, I want to give them some material to work with. So, I'll pose this final question to all of you. What is the one concrete action, commitment, or cooperation mechanism that's achievable by 2027 that you would put on this list for the co-chairs to bridge AI divides before we reconvene in New York. So, I think, Ambassador, since you last had the mic, I'll start with you, then Alessandra. We'll come back down this way and end with you, Honorable Minister.

Thank you. Yes, I think capacity building is not new for the UN. Just remember the General Assembly resolution on AI capacity building. Sometimes it's a matter of implementation because the ideas and the examples of what should be done, they are already there. It's a matter of implementing them. But I think a good example is the Global Network of Centers for Exchange and Cooperation on AI Capacity Building. It was just established yesterday, if I'm not mistaken. Because this is a voluntary mechanism where institutions, universities, AI labs, research centers, they collaborate and pool resources and share available capacities, including courses and skill development. We have 2 Brazilian institutions joining this network. And, you know, I remember in the past people would talk about a CERN for AI because we are in Geneva. CERN is not far away, but as a facility that would represent transnational and multinational collaboration. So I think it's something that we should explore even more. But the last point I wish to make to conclude is that we need to connect the Bletchley process with the UN Global Dialogue. Because you, you know that Bletchley had the, the first AI Safety Summit, then Seoul, the Paris summits. The last one was in New Delhi, AI Impact Summits. They will— and the next will be in 2027 here in Switzerland. But the Bletchley process has no connection with this global dialogue. So I think to avoid duplication and streamline initiatives without creating yet another UN process, my suggestion is that for the next summits, try to connect Geneva and New York so that we can avoid the idea that this is just a standalone event with no follow-up

Yes, I'll keep it short. Investment for upskilling, for literacy, women, kids— we need to bring them forward, we need to bring them along. Not vanilla education. It's not to get an answer out of those AI systems, it's reinvent the process, is to create value, and value is not a shortcut to an answer. So investment in education to support the civil society. Thank you.

Thank you. Varvashi?

I guess it's— we need to also be honest about capacity building. It's going to take a long time, and we don't necessarily have that kind of time, especially when we start thinking about this at multiple layers of government. So I think the question for us is not just, you know, how do we train more officials and how do we build more capacity in government, but what else works? And I think the answer lies in an ecosystem approach. So don't just fund individual capacity-building programs for government or so on and so forth. Fund the ecosystem, the local research organizations, the civic tech groups, the professional bodies in different regions, in different contexts. These are the institutions that persist between governments, between political parties. And this capability then lives in that ecosystem, not in these one-off training programs. And the second thing that I would say is that we need to support the development of open and local evaluation infrastructure that is easy to use by non-technical folks. And the reason I say that is because that is the way that local communities can participate in shaping the values and the goals of AI systems. And that's the ultimate capability Dependency is inevitable. I think we should not assume that dependency is something that only exists in the AI ecosystem. We see it in all sectors. Dependency is not the problem. The problem is, do you have the agency and the power to be able to manage that dependency on your own terms? And I think building open local evaluation infrastructure gives you a really important lever to be able to create a situation of managed and equitable dependency. Thanks.

Okay, thank you. Yeah, so I sit at the University of Pretoria as Director of the African Institute for Data Science and AI, and we're always training the next generation of researchers in AI, also bringing together other researchers, professors in these spaces. The one challenge that you're seeing now as you're building up the capacity, especially at that level, is then you have have to then go and make peace with understanding how are we going to work from brain drain to brain circulation on our continent, right? The drain being that with Africa having the young population that it has, these future skills that people are looking for end up having to come from places like the continent, and as such, we must plan that, yes, people might leave, they might have mobility, but then how do we plan that they also add to the available capacity even when they are somewhere else in the world? On the other side, to keep this going, we need to have opportunities for people to be able to do deep work and deep innovation on the continent. As was said about thinking about an ecosystem, on one part. I also have to wear a hat as a startup founder at Lilapa AI, and the reason we started it— Lilapa means home in Setswana, my mother's language, right? Having a home for researchers to stay on the continent and work on some of our deepest challenges that are having to do with AI. We work on African languages, but it also requires the private sector to start investing, whether it's 0.5%, 1%, of their profits after tax in research and development. If you're not doing that, you're not preparing yourself for the future. All it is going to be is just consumption, and then seeing that most of the benefits of new technology or innovation keeps on accruing not in the global majority, but in very small hands. To spread that, we need more activity locally, and then creating jobs locally, having vibrant ecosystem, And that's how I look at what it means to have sovereign AI. Thank you.

Thank you. In fact, thank you. Thank you. Thank you. In fact, the keyword here is sustainability because many countries have— and here we are talking about also capacity building once again because we've seen a lot of countries that I have seen capacity building projects where they spend money and have seen like a report, uh, training small groups of people and then having a report again, but it lasts just temporary. For AI is something completely different. It has to be sustainable. So we need to build cooperation mechanism and build institution, not dependency as we are today. So for this, we need to blend financing, public-private investment and partnership, and South-South cooperation. And last but not least, the real test for international cooperation today is not whether we produce more declarations, as we did in the past, but how we test whether countries leave this process with stronger institutions, stronger skills, stronger digital foundations, and more capacity capacity to govern this AI themselves. And last but not least, again, sorry, I would like to— I don't want to be remiss if I don't recognize the presence of His Excellency Mr. Antonio Guterres, UN Secretary-General, and I take this opportunity to express our deepest gratitude and appreciation to him for all his efforts in favor of humanity. Including the subject we are discussing today, and please accept President Abdelmajid Taboun's regards and full support to you here and to all the presidents. Thank you.

So thank you all for your time today and for the wonderful contributions from our panelists. Thank you, Your Excellency, Excellency, Secretary-General, for making time in your busy schedule. I can imagine the number of places you could be, but you decided to join us for this important discussion, so we are very appreciative. You know, as I conclude, if I can just leave you with one takeaway from this cluster, I would say let it be this: that capability is the new capacity. And not just building capacity to use AI, but really investing in governments and the local ecosystems, from the grassroots organizations, and really the entire ecosystem so we can build— we can build it sustainably, but really give people agency and the power to shape it in their context. So please join me again in thanking our panelists and for your time today. Thank you.

Well, I would like to thank Ms. Crystal Rugegi for her moderation and really thank all of the panelists today for the excellent discussion that took place, reminding us to think critically when it comes to AI in our workforces, to engage young people, to foster national solutions, and to close the oversight gap. It is now my pleasure to again invite our co-chairs, His Excellency Mr. Samba Diop and Mr. Jovan Kurbelija to moderate a second round of interventions from the floor. So I invite you to please come again to the podium.

Thank you very much. A lot of exercise today. It's almost healthy diplomacy. and gym diplomacy, you know. Thank you for organizers doing it. It is innovation in diplomacy, you see, and you make a lot of moves. Thank you. Okay, I guess, Samba, you share my view about health diplomacy. Good. Our next speaker is His Excellency Angel Banjov, Ambassador of Bulgaria to the United Nations. Ambassador? I can't see. Then Ambassador of Nepal. No, we go with the same stakeholder group? Yes. Okay. Now then we have Ambassador of Nepal, His Excellency Ram Prasad Subedi, Permanent Representative of Nepal in Geneva. Ambassador, please.

Thank you, Mr. Co-Chair. I'd like to begin by the panel's report that provides a baseline for our discussion. AI capacities are highly concentrated in a few countries and firms. Majority of the countries of Global South are outside the global AI development, and they have always remained to the side of consumer, though we have been talking about mainstreaming them as a co-developer. And this AI divide is about capacity to influence. And it's about infrastructure, it's about talent and a governance capability. For Nepal, this dialogue is especially important as a landlocked mountainous country and a graduating LDC. We continue to face real challenges in connectivity, infrastructure, affordability, digital literacy, and access to reliable data and computing resources. These gaps are even more pronounced in rural and remote areas, where communities often remain beyond the reach of basic digital services. Bridging AI divide therefore remains a strong challenge. Our National AI Policy 2025 keeps digital digital and AI to be a human-centric and ethical lever to develop the country. Therefore, what is that we need the most is to build local expertise, strengthen institutions, develop AI applications, and address the national priorities, including disaster risk reduction, agriculture, education, health, and public service delivery. Therefore, what we need is the open interoperable standard and, and thriving innovation ecosystem. Therefore, we support the Secretary-General's proposal of the Global Fund on AI as a pooling mechanism for bridging AI divides, and we believe that this Fund will accommodate the specific needs and situations and contexts of the LDCs and developing countries and LLDCs. We have been discussing about the gap, discussing about investment and cooperation, and therefore I leave you a question: what can be the appropriate way to do it. Thank you.

Thank you, Ambassador. We will have to answer this question, and it's a difficult one but timely and important. We have just one announcement, and I hope that our interpreters, who are doing very hard work, will accept our plea from the participants to stay till 6:15, and we'll then continue in English. Can I see the wave from the interpreters? Yes, of course, sir. Of course, of course, of course, sir. Of course. Therefore, 6:15, then, then we continue after 6:15 for 15 minutes in English. We have still a long list, and it's important that we hear all voices in this session. Our next speaker is Tatjana Galdas-Lottinger, founder of the International Women X in Business in Ethical AI. Tatiana, is she in the room? No, I cannot see her. And then the next speaker is the Executive Director of the UNITAR, United Nations Institute for Training and Research. Nope. Well, we may even finish faster. And then we will have from stakeholder list, let me see, Stakeholder, stakeholder. Well, that's the Ambassador of Botswana. No, I'm sorry, Minister of Communication Innovation of Botswana, His Excellency David Tsere. Tsere, please, Minister.

Yes, I'm here, Chair. Thank you very much for the opportunity. Well, I gave my speech earlier on talking specifically about how developing countries are left behind and how this dialogue should formulate ways on how to assist developing countries. And I hear the message coming out here saying developing countries should partner with other countries or organizations to basically figure it out, but we still believe that this dialogue could perhaps come up with an idea to have a global AI fund for developing countries so that we make a deliberate attempt to get everybody on board, including developing countries, because, like we mentioned, the major issue is the infrastructure development in Africa, and like speakers who spoke here, They highlighted how, if you don't have that infrastructure, you cannot really harness the benefit of AI. So this is still a challenge, and I think we should move just beyond recommending that we should partner with developing countries and come up with a specific and more actionable item to assist developing countries. Thank you very much.

Thank you. The next speaker is Her Excellency Sara Maria from Philippines.

Thank you, Chairperson, Excellencies, distinguished delegates, ladies and gentlemen. The Philippines welcomes this dialogue on artificial intelligence governance and commends the United Nations for convening member states on a matter that now bears directly on development, public institutions, economic participation, and distribution of opportunities across and within nations. Artificial intelligence offers significant possibilities for innovation, public service delivery, productivity, education, health, disaster preparedness, and sustainable development. Yet these possibilities are not available to all countries on equal terms. For developing countries such as the Philippines, the question is not only how artificial intelligence should be governed. It is also whether countries have the digital foundations, institutional capacity, and human capital necessary to participate meaningfully in the AI economy. The AI divide is therefore not only a technological gap. It reflects differences in connectivity, Computing capacity, quality data, technical expertise, research systems, financing, and institutional readiness. Countries with advanced digital infrastructure and mature innovation ecosystems are better positioned to benefit from AI. Countries still building these foundations face the risk of being placed further behind, not because of lack of ambition, but because of unequal starting conditions. The Philippines continues to advance its digital transformation agenda through efforts to expand connectivity, strengthen digital government systems, promote digital literacy, and prepare the workforce for a technology-driven economy. Human capital development is central to this work. Education systems, training institutions, public agencies, enterprises, and workers must be supported as technology changes the organization of work and the demand for skills. Digital literacy, technical training, workforce reskilling, upskilling, and lifelong learning must therefore form part of any serious strategy for bridging AI divides. These efforts must also reach beyond major urban centers. Women, youth, persons with disabilities, Indigenous peoples, workers in transition, small enterprises, rural communities, and geographically isolated areas must be included in the design and implementation of capacity-building programs. Inclusion cannot be limited to access. It must include ability to use, question, improve, and govern technology. The Philippines reaffirms its commitment to international cooperation in shaping safe, secure, trustworthy, and development-oriented artificial intelligence governance. Thank you very much for your time.

Thank you. Thank you so much, Maria. The next speaker is from Peru, Diego Belevan.

Gracias. Buenas tardes. Thank you and good afternoon, distinguished colleagues, participants. In the current context, We consider that this cluster for us is of the utmost priority for dialogue. And allow me to explain why. There is a simple reason. Without real capacity, the principles that we might eventually agree upon simply cannot materialize. The problem for many countries is not the absence of shared values, but rather the impossibility of operationalizing them. And therefore, cooperation is of its greatest value when it complements the normative layer with concrete access to infrastructure, tools, information, and technical assistance addressing priorities and needs expressed by each country. The logic should be of reciprocal benefit and horizontal cooperation, not unidirectional transfer of predefined models. And here, allow me to share 2 elements from our national experience. The first is the national strategy of the government, 2026-2030, which recognizes data as a strategic asset and has 3 pillars: regulatory framework, interoperability, and affordability. Then, in terms of our participation in the working group on governance and data, IT for Development of the United Nations, together with other countries from our region. Our policies around data have been recognized as points of reference for the region. Open source, open models, and open AI could significantly contribute to bridging divides, provided that their adoption is accompanied with developing institutional capacity. This has been recognized. On the contrary, we risk transferring the gap from access to a gap for use. Finally, I'd like to highlight the importance of training in the public sector as a cross-cutting issue, mainly because of its multiplier effect. In many countries, states have services that want to implement these systems lack the necessary tools and knowledge to implement them in their institutions. Addressing that gap is as urgent as addressing the infrastructure gap. Thank you.

Thank you, Director, Director Belevan. Our next speaker is Her Excellency Sina Lawson, Minister of Digital Economy, Transformation of Minister? No, she's not here. And after her, His Excellency Carlos Mendoza Alvarado, Secretary of Planning and Programming of the Presidency from Guatemala, please.

Thank you. I will speak in Spanish. Moderators, distinguished delegates, Guatemala considers that closing gaps in AI is not a secondary issue. It is the very basic condition to ensure that these technologies can contribute to sustainable development and yield shared benefits. When we speak about AI, often we think firstly in models, algorithms, and technological capacities. But for countries like Guatemala, the starting point is different: connectivity, data, digital infrastructure, institutional capacities, and human talent. There cannot be a true democratization of AI if digital divides continue to be reproduced. And reflect economic, social, and territorial divides that already exist. Therefore, for Guatemala, the development of capacities is not limited to training specialists. It also involves strengthening the state, improving scientific research, improving digital literacy, and generating the conditions so that each country can develop, adapt, and use AI solutions according to their own needs and their own priorities. From the Planning Secretariat, we are deploying efforts in order to strengthen information systems for planning, monitoring, and evaluation of development. For us, data is not simply a technical input. It's a strategic asset for better public decision-making and for better orientation of state action. The update of our national development plan, Khartoum 2052, offers a concrete opportunity to incorporate digital transformation and AI into a national development agenda with— which is territorial, inclusive, and sustainable in focus. Guatemala reiterates its commitment to international cooperation, the exchange of knowledge, and the development of capacities as pillars to build an ecosystem which is fairer and more inclusive in AI, because bridging divides today is the necessary condition to ensure that AI can expand opportunities tomorrow. Thank you.

Thank you, Excellency, for bringing in particular non-technical aspect of AI developments, which is— which has been echoing in today's discussion. Our next speaker is His Excellency, Undersecretary from the Ministry of Transport and Communication Information Technology of Oman. Please, Excellency, go ahead.

Thank you. Excellencies, distinguished guests, the promise of AI will mean little if it reaches only some nations and leaves others behind. Today, the divide is real. It is a divide in model development, data, skills, and in access to computing infrastructure and power. Left unattended, AI could widen the gap between nations rather than closing it. For Oman, through the National AI Program, the first foundation to bridge the divide is talent. No nation can benefit from a technology it cannot use or build. This is why we have invested heavily in talents, and skill development through several national initiatives focused on AI product management, data science, chip design, cloud services, and others. The second foundation is AI infrastructure. Oman is developing a national digital triangle to develop interconnected multi-gigawatt digital hubs to host green AI and high-performance computing clusters. Building on Oman's political neutrality, abundance of submarine cables, and green energy diversity, the Digital Triangle is a safe space to host and process the next generation of trusted intelligence. The third foundation is cooperation. No country can cross this divide alone. We emphasize on the importance of knowledge sharing, and international collaboration so that every nation equally benefits from AI. Excellencies, Oman stands ready to play its role as a partner and as a bridge. We are willing to share what we have learned, work on joint projects, and cooperate with the United Nations and member states so that no nation left standing on the far side of this divide. Thank you.

Thank you, thank you so much. The next speaker is from Pakistan, Minister of ICT.

Excellencies Chair, distinguished co-chairs, colleagues, a very good evening. As artificial intelligence reshapes power faster than governance can adapt, allow me to frame the divide this cluster is really about. It is not a divide in access to tools. Rather, it is an asymmetry in the ability to shape the rules, the standards, and the foundational infrastructure of this technology. Our multipolar world is becoming increasingly technopolar, where power concentrates around those who control frontier compute, models, and data. When you look at what is at stake of this asymmetry, it is not only a development concern, it is a question of global stability. If we do not take consequential decisions, the trajectory is clear: 2 parallel AI ecosystems and a permanent 2-tier order in which most of the world's nations are structurally locked into the role of rule takers. In the interest of global stability, therefore, I want to offer a shift in the frame. We should aspire to move towards a model of polycentric architecture of co-ownership. The lessons from other complex global challenges is that governance works not through a single institution issuing solutions from the top, but through many overlapping initiatives orchestrated so that they reinforce rather than fragment. This dialogue and its scientific panel can be that orchestrator, a smart conveyor belt that turns scattered voluntary efforts and best practices into durable shared architecture. And the emerging economies must be at the center of the design table for open foundational capabilities— the models, the datasets, the compute access frameworks, the evaluation benchmarks, and the safety regimes. I call for an end to the you decide, we receive era. As co-architects of what we build together, we will build an inclusive architecture that respects our cultures, our singularities, and fully utilizes our strengths. This is why interoperability and strategic openness are non-negotiable. Open standards, open data, and open models are the fair strategic equalizer that lets nations leapfrog while preserving their sovereignty and their right to choose. Pakistan is already architecting this principle. We have set our own national AI direction and invested in building capabilities across our public sector, and we are ready to contribute our open-source-powered stack. We are addressing real-world use cases in key economic sectors, climate resilience, agriculture, and inclusive public service delivery and multilingual data. We have a willingness to experiment openly and govern responsibly. in service of the global commons. So let this dialogue catalyze the concrete mechanisms of co-ownership with a shared future co-created with emerging economies. Frontier economies are the drivers of the world economic growth. We will also be architects of its AI foundations that bring prosperity to all. I thank you.

Thank you so much. The next speaker is from Namibia. The Minister of ICT, Iman.

Thank you so much, Chair, distinguished guests, ladies and gentlemen. Namibia welcomes this global dialogue on AI, and like many other Global South countries, Namibia is preparing its national AI governance framework, taking into consideration the following: human-centered governance, targeted public sector innovation, public-private and global collaboration, equitable infrastructure, and the globally accepted interoperable rules and shared standards. Yet these areas mean nothing if we don't address the following now rather than later. First, financing and investment, unlocking the financing to strengthen digital infrastructure in developing states. We heard earlier from UNCTAD that only in this financial year alone, $800 billion spending for AI will definitely not go towards developing countries. Secondly, high technical skills, academia, especially training agreements between universities in the Global North and Global South to train as many young people across the Global South as possible to add to the capacity in that area. Third, access includes lean models, open standards, and greater accountability geared towards mega-tech companies, especially where matters of technologically facilitated gender-based violence is concerned. Ladies and gentlemen, just as electricity evolved from a dangerous, mysterious force into a universal background utility that powers everything from healthcare to agriculture, AI must be shaped to elevate human potential rather than replace our core judgment. AI should perform the mundane tasks in order to free up human capacity to innovate, practice empathy, and unleash our creativity. That is the real test for capacity building for AI. Finally, Namibia welcomes the 2nd Global Dialogue set for May 2027 in New York because global cooperation is strengthened through continued dialogue. Thank you very much.

Thank you, Minister, for highlighting again empathy and other aspects of digital development. I can see that we are getting more and more human inputs and less and less algorithm and technology, which is an interesting echoing tone of discussion. Our next speaker is His Excellency, Minister of Digital Transformation of Ivory Coast. Let's see if minister could raise his hand. No, no. And next is His Excellency Permanent Representative of China to the United Nations Ambassador, please.

Thank you very much, co-chairs. Currently, AI is profoundly reshaping the development landscape of of countries and the global governance architecture. We should take universal benefit and inclusiveness as our starting point to uphold the role of the United Nations as the main channel and focus on capacity building to shape an inclusive, open, sustainable, just, secure, and reliable digital intelligence future. China has the following suggestions. Firstly, uphold equitable development and bridge the digital and AI divides. We must abandon the mindset of technological hegemony and barriers and endeavor to narrow the development gap amongst countries, regions, and groups. We should create an open, fair, and non-discriminatory environment for AI development, continuously improving its inclusiveness and accessibility. Digital sovereignty means that countries have the right to independently choose AI products without being coerced into taking sides. Second, we should promote innovative development while guarding the red line of security. We should uphold the spirit of openness and sharing, promote the continuous emergence of innovative achievements and the accelerated expansion of applications and scenarios, thereby enhancing the level of AI innovation and development globally. Meanwhile, we should ensure both development and security, maintaining the steady and long-term development of AI with solid security safeguards. Thirdly, we should strengthen integrated development and improve global governance. We should accelerate the construction of digital infrastructure and improve the layout of globally interoperable AI and digital infrastructure. Countries should proactively strengthen policy and regulatory coordination, oppose drawing ideological lines, and develop global AI governance frameworks, norms, and standards based on broad consensus. China has consistently been an advocate, promoter, and a pioneer in AI capacity building. China proposed the Global AI Governance Initiative and the AI Capacity Building Action Plan for Good. And for All, which facilitated the adoption of the world's first UN General Assembly resolution focusing on AI capacity building, jointly initiated with the Group of Friends for the International Cooperation on AI Capacity Building, which was launched with Zambia, and successfully held workshops on AI capacity building on the subject, taking concrete actions to help countries share the benefits of AI. Now, from July 17th to 20th, China will host the 2026 World Artificial Intelligence Conference and a high-level meeting on global AI governance in Shanghai. We look forward to the active participation of all parties, deepening exchanges and cooperation, and creating a better future together. Thank you very much.

Thank you, Ambassador, for also announcement of this important event in Shanghai later on during the month of July. Our next speaker is His Excellency Darkan Zazubkay, Chairman of the Agency of Civil Service Affairs of Kazakhstan.

Thank you. Dear colleagues, today's global discussion focuses mainly on 2 dimensions: how to advance innovation and how to mitigate risks. But there is an important point: is the state ready to govern effectively in the age of AI? You believe that public trust in AI does not begin with algorithms. It begins with trust in government, in how fairly, transparently, and responsibly public steps are 10 years ago, we set, we set the goal of ensuring that personal choice is based and merit. This led to the creation of our digital HR ecosystem, e-Kosmet. Today, e-Kosmet serves more than 80,000 civil servants, supports over 400 digital HR process and is integrated with more than 100 government information systems. Today, Kazakhstan is taking the next step. The year 2026 has been declared the Year of Digitalization and AI. We have adopted the Law on AI and the Digital Code. Our agency is applying AI to assess HR data and moving to predictive planning and the AI audit of government functions. Same time, our position remains clear. AI may assist but never replace human responsibility. Our approach to AI in the public sector is guided by 3 fundamental principles: transparency, human oversight, and responsibility. Dear colleagues, Kazakhstan built its digital foundations step by step. That is why our experiences is now attracting international interest. The X-MED project has been published on the OECD of the platform as innovative public governance practice. The point is not whether AI can transform the state. The point is what values the state embeds in the use of these technologies. Kazakhstan has chosen a way based on transparency and fair opportunity. and we're ready to share our experiences. Thank you for your attention.

Thank you. Thank you, Dahan. Our next speaker is Her Excellency Jessica Hunter from Australia.

Thank you, co-chairs and distinguished delegates. Capacity building is at the heart of Australia's approach to strengthen strengthening global connectivity. We're committed to supporting partners with the skills, tools, and systems needed to connect communities and enable inclusive digital development. We see capacity-building initiatives as essential to building the means and the mechanisms to ensure that the development and deployment of AI is not only widely accessible but also adopted in a trusted manner that respects human rights and delivers benefits equitably. However, there is no one-size-fits-all solution. AI capacity building must be underpinned by a comprehensive understanding of the unique challenges and opportunities in adopting AI across the AI lifecycle, including governance, skills, and digital infrastructure. Australia is supporting AI capacity building programs across the Pacific and Southeast Asia through a focus on skills, governance, and infrastructure. In the Pacific, our efforts focus on strengthening foundational digital capabilities through initiatives such as the Australia-Pacific Digital Economy Program. This sits alongside partnerships with governments to build public sector capacity and apply emerging technologies in other areas such as disaster preparedness, climate monitoring, and service delivery. It is further underpinned by investments in connectivity. In Southeast Asia, Australia works with governments, industry, and academia to advance inclusive digital transformation, including workforce training in AI-related fields, sharing best practices on AI governance, and supporting innovation through collaboration with startups and research institutions. We are also engaged regional organisations such as ASEAN to promote standards alignment and interoperability. Across all of our engagement, Australia is guided by the principles of inclusivity, trust, and safety. Australia is committed to ensuring that AI systems are human rights respecting, safe and secure, and deliver benefits across all segments of society, including to women, children, and linguistically diverse communities. Internationally, we already have strong multi-stakeholder community and established international law to support the realization of maximum benefits for the use of digital technologies such as AI. It is up to us here in this room to ensure that these benefits are available to all. The need to work together is clear. We must now be collaborative in our efforts to bridge the digital divides and ensure that our vision for a free, open, and interoperable digital future is accessible and available to all. Thank you.

Thank you so much. Next is Her Excellency Larissa from Belarus. Larissa is not around. The next is Her Excellency Hoda Baraka from Egypt.

Thank you very much, Chair. This is Ahmed Hefnawi on behalf of Dr. Baraka. So I'll just focus my intervention on 3 very specific areas. Number one, equitable participation, which will not be achieved by closing gaps one at a time— infrastructure, then skills, then institutions— because divides compound rather than sit side by side. Our view is that international cooperation should be structured around sequenced, connected packages. Financing, scaling, and market access mechanisms designed together, not funded separately and hoped to align. The most binding barrier is rarely a single missing input. It is the absence of a pathway from access to genuine participation. Training that does not lead anywhere is not capacity It is a credential. We would like— we would ask this dialogue to evaluate capacity-building proposals against that pathway test, not against inputs delivered. In Egypt, our Digital Egypt Pioneers initiative was deliberately paired with the Industry Development Agency's GIGS program and the freelancer eg platform, now backed by formal freelancer taxation legislation, precisely so that reskilling would connect to real income, not stop at a certificate. The second topic that we wanted to highlight is open systems that can be genuine equalizers, but only under 2 conditions. These dialogues should treat as a package. Not, not a choice. Openness must be paired with governance that keeps trust intact, and it must be matched by domestic capacity to actually use what is opened. An open model without local infrastructure or skills does not close a divide. It simply relocates the barrier from cost to capability. Our view is that international support for OpenAI should therefore always travel with support for the conditions of its use, and that safeguards against misuse— against misuses should be designed alongside openness from the outset, not retrofitted once harms appear. Um, in Egypt, we open-sourced Karnak, which is a national Arabic large language model. While governing its downstream uses. A legal assistant, a language tutor, healthcare tools under our national guidelines for trustworthy and responsible AI.

So openness and safeguards. Sorry. No worries. That's the— but we are— thank you for bringing all of this really rich rich scene, AI scene in Egypt. And we have our last speaker on the list, His Excellency Permanent Representative of Ethiopia. Ethiopia,

please, Ambassador. Thank you, co-chair, distinguished co-chairs, excellencies, distinguished participants. Ethiopia thanks the independent International Scientific Panel on Artificial Intelligence for their report and notes the contents therein. The report in particular recognizes that one of the greatest global AI risks is not only the misuse of artificial intelligence, but also the unequal access to its benefits. For Ethiopia, bridging this divide is a national priority. Through our Digital Ethiopia 2030 strategy, we are investing in the foundations necessary to build an inclusive AI ecosystem. These efforts include the establishment of the Ethiopian Artificial Intelligence Institute, the launch of the 5 Million Ethiopian Coders initiative to develop digital and AI skills, and the planned Artificial Intelligence University to strengthen research and innovation. And the expansion of our Fayda digital national ID system as a key component of digital public infrastructure. We also— we are also promoting AI applications in healthcare, agriculture, education, public service delivery, while supporting the development of AI solutions that reflect our languages and context. Excellencies, No country can bridge the AI divide alone. That is why international cooperation and a genuine multi-stakeholder approach are very important. Governments, academia, innovators, the private sector, international organizations, civil society, and the technical community all have complementary roles to play in expanding access to artificial AI infrastructure. Infrastructure, strengthening capacity, promoting knowledge transfer, and supporting responsible innovation. Developing countries like that of ours continue to face multiple digital and technological divides. However, we also recognize that we are at the early stages of the AI era. This presents a unique opportunity to address these disparities before they become entrenched. Investing today in capacity building, digital public infrastructure, equitable access to computing resources, and inclusive governance will help ensure that AI becomes driver of sustainable development rather than a source of greater inequality. In closing, Ethiopia supports strengthened international collaboration to ensure that every country has the opportunity Bridging the AI divide is not only a matter of fairness, it is essential to ensure that artificial intelligence serves all humanity and contributes to achieving the sustainable

development goals. Thank you, Ambassador. I don't know if there is a coincidence, but you You're the last speaker on the list and it leads me to one message with which I would like to conclude, at least from my side, and then my co-chair will continue. You're coming from the country where allegedly the Ark of Covenant is kept in Aksum at the north of the country and it is one of the earliest codifications of human wisdom. And this is probably the message for us today that knowledge and wisdom crosses across the eras and great civilizations from Asia, Africa, Europe, Latin America. And what we are doing today, we are revisiting this wisdom and knowledge. Therefore, there should be also that sense, historical sense of the discussion, and that knowledge and wisdom which is in core of AI is probably the critical challenge for all of us. Knowledge, what was said today, defines us as humans. Local knowledge, knowledge of indigenous communities, knowledge of the local communities worldwide, and the key challenge for us is to use technology, advanced technology, to ensure that that knowledge of the past is preserved and that future knowledge is developed in inclusive and enabling ways. We heard many excellent examples on capacity building, closing knowledge, and I would say cognitive gaps sometimes, use of open source, and excellent points on AI sovereignty as enabling approach, not necessarily a sort of restricting approach to digital developments, inclusion. And we have a really rich menu which we will codify with some sort of distance of few hours till tomorrow's concluding session, but that would be my first comments based on, let's say, echoing themes in discussion and your excellent intervention. I would like to invite my co-chair and colleague to provide his concluding remarks before we make good news for the interpreters that we'll finish exactly

on time. Thank you. Thank you, Johan. Thank you, all of you guys. We are really very happy to hear the great speech from each one of you. And capacity building is something that we all align, that the most important is how to— how we're going to address it in our different environment. We talk about digital sovereignty. For us, we don't have the same understanding of what is digital sovereignty. For us, it's about upskilling. We don't have technology, but we have people that can understand and execute the technology. The technology is similar, is the same language. We can understand it. The other point is about bridging the gap. We can see where AI is is booming is the big countries where they have demography and they have a big investment. In our countries where I'm from, we don't have such kind of demography first. Second, we cannot afford this big investment. That means what? That means we should collaborate and work together in different countries to build our own AI in our understanding. The other one is the— what I call just now the regional strategy. As small countries, we cannot build our own AI, so we should work together with different countries in order to build a regional strategy for the data governance. I don't go to the technical aspect where there is strength about our LLM model, how to strengthen it, how to improve it. Thank you. I will just finalize by saying AI is shaping our lives today. Tomorrow, AI will drive our journey. So we should be ready and prepared for it. This is my last sentence to close, and thank you very much for all your standing

here. Thank you. Today's session, by thanking our distinguished co-chairs, Mr. Sambadewa and Mr. Jovan Kirbalija, and all of our panelists and speakers. We really heard a clarion call for collaboration and for us to work collaboratively to achieve equitable access to AI for all through inclusive participation. and also support for local innovation. So with this, I thank everybody once again, and I officially close today's session. Thank you. Thank