High-level governmental plenary segment: Harnessing the benefits of AI for all through inclusive and interoperable approaches - part 1

Estonia argues that AI governance should be centred on human welfare rather than technology for its own sake. It says AI must be transparent, accountable, democratic and secure, while regulation should still leave room for socially useful innovation.

The speaker explicitly set out five principles: AI must be human-centred, trustworthy, protective of democracy and the rule of law, secure and resilient by design, and supportive of innovation . He added concrete expectations that people should know when AI is used, how data is handled, and who is responsible for decisions, and that decisions affecting rights must be explainable and accountable .

Estonia presents AI governance as something that should be internationally coordinated through practical cooperation rather than fragmented national approaches. It links openness and democracy to common standards and interoperable rules across like-minded partners.

The Estonian president said directly that the future of AI governance should be 'open, democratic and practical' and called for interoperable rules, shared standards and cooperation among democratic partners .

Estonia frames AI as a tool for practical improvement in public services, education, business and administration. It also argues that these gains must be accompanied by broad societal AI literacy so that everyone can use and understand the technology responsibly.

The speaker gave concrete examples of AI helping doctors spend more time with patients, enabling teachers to support learning, helping entrepreneurs create higher-value products and helping public servants reduce bureaucracy . He also cited Estonia's est.ai initiative and AI Leap programme, which provide AI-based learning tools for students and teachers and promote critical and responsible use .

Estonia argues that inclusion is not only about access to systems but also about the ability to understand them. It treats AI literacy as an equality issue that should reach people across schools, workplaces and public institutions.

The speaker said this was 'a question of equality' and argued that if AI changes every sector, every person must have the chance to understand it . He added that AI literacy must reach widely across society, from schools to workplaces and public institutions .

Estonia stresses that AI security is not a technical add-on but a core democratic and national security requirement. It argues that AI systems must be built to withstand malicious use, cyber threats and operational disruption.

The speaker said Estonia's experience as a digital state had taught that cybersecurity is foundational . He then stated that AI systems must be protected against misuse, manipulation, cyber attacks and disruption, adding that resilience is a democratic value in a world where digital systems are part of national security .

El Salvador supports using AI for development and public benefit, but it rejects any governance model that allows AI to replace human responsibility in rights-sensitive decisions. The argument is that innovation and governance can work together if clear safeguards preserve human oversight.

The vice president said AI must be placed at the service of the common good and the development of peoples . He also stated that AI can strengthen public services but cannot replace responsibility, human judgement and oversight in decisions affecting fundamental rights and guarantees .

El Salvador argues that national governance alone is insufficient and that international collaboration is needed for effective AI governance. It links cooperation to practical outcomes such as capacity-building, sharing experience and reducing exclusion.

The speaker said it was essential to strengthen international cooperation, bridge the digital divide, share good practices, develop technical capacity and ensure that no country remains excluded from AI benefits . He added that AI governance progresses best when it facilitates interoperability, trust and practical cooperation among states, the private sector, academia, civil society and international organisations .

El Salvador presents AI as a broad development tool that can raise productivity and improve lives across key sectors. The argument emphasises practical state use of AI to strengthen services and social welfare.

The vice president listed health, education, agriculture, industry, security, justice and public administration as sectors where AI is opening unprecedented opportunities . He said these applications could strengthen public services, boost productivity and improve quality of life for millions .

Georgia portrays AI as a powerful force that can influence perceptions, narratives and political order, not just a neutral tool. It argues that governance must prevent concentration of power and misuse that undermine liberty, sovereignty and human values.

The president warned that bad-faith use of AI could distort perceptions of reality and trigger global instability . He also said the concentration of AI power must not become an instrument of totalitarian control and new digital tyranny that infringes national sovereignty and individual liberty , and that AI systems must be aligned with human values .

Georgia sees stronger international cooperation as essential because AI’s risks and benefits are global in scope. It specifically links this effort to implementing the Global Digital Compact and building an open and secure digital future.

The speaker said it was essential to elevate international cooperation to a qualitatively new level . He called for laying the foundation for an open, free, secure and inclusive digital future and highlighted the effective implementation of the United Nations Global Digital Compact as particularly important .

Georgia argues that AI can be a major positive force for science and development if it is guided by the right values. It sees the technology as capable of advancing medicine, reducing poverty and helping tackle climate challenges.

The president said AI should become a catalyst for expanding human potential, innovation and human progress . He then gave examples including transforming medicine, biology and quantum physics, overcoming incurable diseases, eradicating poverty and effectively responding to climate challenges .

Georgia emphasises that AI misuse has consequences far beyond technical failure, including destabilising politics and society. It treats these security and existential risks as a central reason for robust governance.

The speaker said misuse or bad-faith use of AI may distort perceptions of reality to such an extent as to trigger global instability . He also warned that losing control over language and culture through AI could place the essence of humanity and the future of humankind at risk .

Luxembourg argues that AI governance must safeguard core democratic and human rights values and cannot rely solely on voluntary norms. Where risks are severe, especially in military contexts, it supports binding international rules.

The prime minister said AI governance must protect human dignity, democracy, human rights and vulnerable groups including children, and that these are collective responsibilities . He cited military AI as a specific example, arguing that human judgement and responsibility must never be surrendered in life-or-death decisions and that binding international rules may be necessary where voluntary principles fall short .

Luxembourg stresses that AI governance is a collective international task and cannot be solved through isolated national action. At the same time, it insists common standards should not erase linguistic and cultural diversity.

The speaker said the choice of what to make of AI cannot be made by any country alone and that multilateralism is not optional but essential . He further argued that common standards should not mean uniformity and that AI must reflect the diversity of languages, cultures and societies .

Luxembourg rejects a homogenising model of AI and argues that governance should preserve cultural and linguistic plurality. It presents diversity as a value that standards must protect rather than suppress.

The prime minister said common standards do not mean uniformity . He explicitly stated that AI should not make the world uniform and should instead reflect the diversity of languages, cultures and societies .

Luxembourg identifies military uses of AI as a particularly high-risk area where human control is indispensable. It argues that life-and-death decisions must remain under human judgement and responsibility.

The speaker used military AI as a concrete example and said that as AI integrates into defence systems and warfare, human judgement and responsibility must never be surrendered in decisions over life or death .

Eswatini argues that AI should remain under clear human authority and should never regulate itself. It especially stresses protecting children and creating enforceable rules rather than relying only on voluntary commitments.

The prime minister said AI systems must not self-regulate and that human judgement, responsibility and oversight must remain at the centre . He also said children must be protected from manipulation, exploitation, harmful content and addictive design, and that legal and governance frameworks must be enforceable nationally and supranationally because voluntary principles are not enough .

Eswatini maintains that responsible AI must support diverse local realities rather than impose one dominant model of progress. This is particularly important for small developing states that otherwise risk becoming passive recipients of systems built elsewhere.

The speaker said AI must not narrow the world into a single understanding of progress, identity or values . He added that AI tools, data and development pathways must be infused with cultural sensitivity and respect for languages, social contexts and normative traditions, especially for small developing states .

Eswatini frames AI risk management as requiring legal safeguards that can be audited and enforced. It highlights privacy, cybersecurity and infrastructure protection as areas where international cooperation and enforceable obligations are necessary.

The prime minister said legal and governance frameworks must provide auditable rules, accessible remedies, assigned responsibilities and cross-border cooperation . He then specified stronger safeguards for privacy, cybersecurity, data protection and the integrity of critical infrastructure .

Poland presents AI governance as a balancing exercise between protecting people and enabling development. It supports a model in which human-centred safety, rights and privacy protections coexist with innovation and economic growth.

The deputy prime minister said AI must be human-centric and safe in real-world use . He added that it should protect rights and freedoms, ensure privacy and security, while also enabling innovation and economic growth .

Poland emphasises practical deployment of AI in public-facing sectors rather than discussing governance only in abstract terms. It links these uses to improved efficiency, simpler procedures and broader access to expertise.

The speaker stated that in Poland the priority is the application of AI . He gave healthcare and administration as examples, saying AI is being deployed there to improve efficiency, simplify procedures and widen expert access .

Poland argues that trustworthy AI governance must address misuse risks directly, especially misinformation and harmful abuse. It therefore links standards and transparency to risk mitigation measures.

The deputy prime minister said Poland addresses risks through trust standards, transparency and measures against disinformation and harm from abuse .

Kazakhstan argues that effective AI governance needs a strong legal and institutional base, including digital rights and technical regulation. It links these safeguards to accountability and compatibility across governance systems.

The minister said a new constitution guarantees citizens' digital rights, including personal data protection . He also noted that Kazakhstan adopted a digital code, an AI law and a cybersecurity law that regulate ethical standards and the labelling of AI products , while also stressing the need for compatibility around data protection, safety, transparency and accountability .

Kazakhstan sees regional cooperation and experimentation platforms as a practical way to advance interoperable AI governance. It offers UN-backed centres as spaces where member states can test and scale digital solutions together.

The minister highlighted two UN-backed regional platforms in Almaty and invited member states to second national experts to them . He described the centre as a regional sandbox for testing and scaling best digital solutions and later said basic approaches should be compatible and that Kazakhstan is ready to share experience, launch joint pilots and develop interoperable approaches .

Kazakhstan argues that AI transformation depends on broad public digital infrastructure and education, not only AI models. It presents its digital public services, connectivity, compute infrastructure and education reforms as the basis of an AI-enabled state.

The speaker said Kazakhstan aims to become a fully digital nation within three years and noted that 92% of public services are online, 90% of transactions are cashless and all personal documents are digitised . He also cited broadband expansion to over 90% of villages, the Trans-Caspian fibre corridor, two supercomputer clusters and embedding AI across education, including in secondary schools and a new AI research university .

Lesotho argues that AI governance should rely on established institutions and be especially careful where systems affect rights or public services. In those cases, human oversight, transparency and remedies must remain central.

The deputy prime minister listed sector-led and risk-based governance through existing regulatory institutions as a priority . She said that where AI affects fundamental rights or the delivery of public services, meaningful human oversight, transparency and effective mechanisms for accountability and redress must remain central .

Lesotho says the key challenge for small and developing states is not just obtaining AI tools but building the capacity to shape and govern them. It therefore prioritises investment in skills, institutions, data systems and infrastructure.

The speaker said the central challenge for small and developing states is the capacity to participate meaningfully in AI development, governance and application . She explained that this requires investment in human capital, data systems, institutions and digital infrastructure .

Lesotho promotes openness as a practical strategy to reduce dependency and enable locally adapted AI development. It pairs this with interoperable digital public infrastructure as the operational basis for responsible public-sector AI.

The speaker identified openness as a core priority and said open source software, open data, open standards and open models can reduce dependency, lower costs, foster innovation and help countries adapt AI to local contexts and development priorities . She also highlighted interoperability and digital public infrastructure, including digital identity, e-government services, secure data exchange and cybersecurity frameworks .

Lesotho links inclusion to the ability to use open technological resources with locally relevant data. Its argument is that countries need data that is organised and prepared for responsible AI use in their own contexts.

The deputy prime minister said open models and related resources can help countries adopt AI solutions suited to their local context and development priorities . She also proposed a framework for AI-ready data, arguing that many African countries do not lack data entirely but lack data that is organised, interoperable and prepared for responsible AI use .

Lesotho argues that open technological resources are a practical tool for inclusion and domestic innovation. It sees openness as a way to reduce both financial barriers and dependence on external providers.

The speaker said open source software, open data, open standards and open models can reduce dependency, lower costs, foster innovation and enable countries to adopt AI solutions for their local needs .

Qatar argues that trust in AI depends on grounding its governance in legal and rights-based principles. It sees safety, security and respect for international law as essential foundations rather than optional additions.

The minister identified respect for human rights as a fundamental principle and said this is a matter of trust . He also stated that international law must be respected in the use of AI and called for safe and secure models .

Qatar argues that AI risks do not stop at borders, so governance arrangements for different models need to work together. It treats interoperability as a practical requirement for coherent risk management.

The minister said the dangers of insecure models do not know borders . He then called for interoperability capacity between different model governance structures so that decision-makers and developers are not exposed to those risks .

Qatar argues that inclusion in AI requires access to the key inputs of AI development and use, not merely end-user consumption. It specifically stresses linguistic inclusion through Arabic-language tools.

The minister said that computing power, data power and model development are concentrated in many countries and that access must be guaranteed to calculation models and open-source available tools . He specifically added that this should include tools in the Arabic language .

Algeria argues that AI governance must confront the real harms of unequal power and inadequate implementation, not just produce abstract principles. It links responsible governance to reducing dependency, inequality and disinformation while protecting rights.

The minister said AI can support sectors such as healthcare, education and agriculture, but without inclusive governance it can deepen inequality, increase dependency, amplify disinformation and undermine rights . He also identified the limited practical impact of existing efforts as a key concern and said principles are necessary but not sufficient without mechanisms and measurable progress .

Algeria sees fragmented governance efforts as a weakness and identifies the UN as the proper venue for legitimate and inclusive international coordination. It therefore supports a globally shared governance framework centred on the UN.

The minister said AI governance must be shared internationally, responsibly and anchored in dialogue, inclusion and legitimacy . He also identified fragmentation of international efforts as one of three main concerns and said this confirmed the United Nations as the most legitimate and inclusive platform .

Algeria warns that the absence of inclusive governance can turn AI into a source of social and political harm. It specifically highlights dependency and disinformation as key risks.

The speaker said that without inclusive governance AI can deepen inequality, increase dependency, amplify disinformation and undermine rights .

Switzerland argues that AI governance must be anchored in international law and shaped through inclusive processes involving multiple stakeholders. It presents human rights and human-centred design as non-negotiable foundations.

The federal councillor said, speaking also for the Freedom Online Coalition, that AI governance must be firmly enshrined in international human rights and international law and elaborated through genuine multi-stakeholder and human-centred processes .

Switzerland argues that the UN should not only produce norms but also model how AI can be used responsibly in practice. It links this to innovation in methods of work and cooperation across the international system.

The speaker said the United Nations should not only define the rules of AI governance but also show an example in practical application . He added that this should happen through methods of work, new forms of cooperation and the accomplishment of the UN's mandate for member states and populations .

Congo argues that AI’s promise must be matched by governance that protects human values and social inclusion. It therefore links innovation to ethical safeguards, data protection and resilience against disinformation and exclusion.

The minister said AI offers remarkable opportunities but also raises legitimate challenges regarding ethics, security, data protection, disinformation and the risk of exclusion . He added that efforts must converge on both innovation and preserving human values .

Congo presents AI as a development tool that can help achieve both global and African regional goals. It links this vision to domestic institution-building, research and training capacity.

The minister said AI could become a major transformative vector for accelerating the SDGs, the African Union agenda and access to essential services . He also described national steps including a ministry for the digital economy, a focus on capacity building and training, and the creation of an African research centre for AI at Denis Sassou Nguesso University to train researchers and innovators .

Guatemala argues that AI governance needs both normative and practical foundations: ethics, rights, evidence and cooperation. It also insists that governance is incomplete if it excludes culturally and linguistically diverse populations.

The representative said AI governance should combine ethical principles, human rights, independently verified scientific evidence, effective international cooperation and concrete resources to bridge divides . He also stressed that a truly inclusive AI cannot be limited to access to technologies designed elsewhere and must recognise cultural and linguistic diversity .

Guatemala broadens the meaning of inclusion beyond connectivity or access to devices. It argues that AI systems should respect indigenous languages, ancestral knowledge and territorial realities, rather than render them invisible.

The speaker described Guatemala as a pluricultural, multilingual and multi-ethnic country . He said digital inclusion must also mean cultural, linguistic and territorial inclusion, including progressive incorporation of indigenous languages, respect for ancestral and community knowledge, and avoiding biases that erase peoples and territories .

The UAE argues that AI governance should actively prevent concentrated and abusive power by making access broader and governance more decentralised. It also sees international institutions such as the UN as necessary custodians of this process.

The minister said that because AI can now make decisions, there is a burden to shape technology to be merciful and avoid tyranny . He then identified three governance ideas drawn from history: separation of power to ensure access for the many, decentralisation through investment in AI infrastructure around the world, and an independent judiciary so institutions like the UN can shape AI's future, followed by the protection of people .

Spain argues that AI governance is a matter of defending democratic society and human rights against serious technological abuse. It supports strong institutions and legal oversight to protect minors, rights and information integrity.

The minister said Spain had expanded its AI law to prohibit sexual deepfakes and protect minors . He also referred to a supervisory agency for AI, a charter and observatory of digital rights, and concluded that AI governance is about defending human rights against 'techno-fascism' .

Spain presents AI as a fast-moving source of systemic risk across multiple domains. It argues that governments have only a narrow window to establish governance capable of mitigating these dangers.

The minister said the speed of AI is multiplying cyber, biological, democratic and security risks . He added that science shows governments have a very narrow window of opportunity and very little time to redirect AI toward beneficial goals through interoperability and governance .

China argues for a global governance model that does not force a trade-off between AI development and safety. It supports inclusive and effective governance tools such as ethics review, frameworks, standards and norms.

The minister said President Xi Jinping had proposed a global AI governance initiative calling for inclusive, equitable and effective governance to ensure AI for good . He later said China had put forward a Global AI Governance Action Plan that upholds balancing development and safety and promotes technical ethics review and other governance proposals .

China argues that AI benefits should not be limited to the most powerful states. It supports international cooperation specifically aimed at helping Global South countries develop capabilities and share in AI gains.

The minister said all countries, regardless of size, strength or social system, have an equal right to develop and benefit from AI . He cited China's AI Capacity Building Action Plan for Good and for All and concrete South-South and triangular cooperation projects, adding that China supports other countries, especially in the Global South, in developing AI capabilities and sharing the AI dividend .

China sees the UN as the central forum for legitimate AI governance and advocates consensus-based international rulemaking. It links this to standards, norms and governance frameworks developed collectively.

The minister said China has consistently supported the UN as the primary channel for global AI governance . He also said China stands ready to work with all parties to advance global AI governance frameworks, standards and norms based on broad international consensus .

China presents open source AI as a shared global asset that can reduce barriers to participation. It argues that accessible open models can promote inclusion, innovation and broad international adoption.

The minister said open source AI is a shared asset for all humanity . He gave Chinese open source models such as DeepSeek and Quang as examples that have significantly lowered the barriers and costs of AI adoption and said China is committed to promoting open source AI through international cooperation .

Cambodia argues that AI governance should not be rigid or purely reactive. It should scale in proportion to actual risk, anticipate future challenges and include the private sector, academia, civil society and citizens as operational partners.

The minister said meaningful adoption requires trust and that Cambodia believes AI governance should be risk-based, responsive and future-oriented, adapting as technology evolves and anticipating challenges before they materialise . He added that Cambodia has adopted a principle-based approach in line with ASEAN and sees the private sector, academia, civil society and citizens as partners rather than mere subjects of regulation .

Cambodia argues that no country can govern AI in isolation and that practical international cooperation is needed. It focuses on helping regulators learn, share knowledge and build common understanding from real-world incidents.

The minister said no country can govern this technology alone . He then identified three concrete areas for international support: building the technical capacity of regulators, fostering exchange of information on emerging AI capabilities, products and services, and strengthening cooperation on regulatory experience and incident monitoring so lessons in one jurisdiction can benefit all .

The participant argues that AI governance is incomplete if it ignores local and regional governments, because many real-world effects are felt in cities and communities. He says AI should be designed for quality of life and resilience, not just technical optimisation, and should safeguard rights where it is actually implemented.

The speaker from the Global Cities Hub argued that AI is used and implemented at the local level and that local and regional governments play an important role in how AI is used . He recommended that AI should not be deployed solely to maximise efficiency or technological performance but should improve citizens' quality of life, and that governance should be designed for resilience and should strengthen public institutions, especially in urban environments .

The participant argues that AI governance must account for how AI changes human cognition and behaviour, not only technical or administrative rules. As people trust AI outputs more, their ability to act as the final check on errors may weaken.

The speaker described a neuroscience-based study in a Swiss company measuring brain activity during workplace tasks with and without AI . He reported findings including a 500% increase in familiarity bias, reduced vigilance as users trusted AI outputs more, reduced scrutiny with higher AI literacy and attention drift as AI use increased, arguing these cognitive effects undermine the human last-mile governance layer .

The participant argues that AI capability does not have to be monopolised by large states or companies. Smaller countries can build relevant AI systems by collaborating, sharing code, and pooling data, infrastructure and expertise.

The speaker from the Kyrgyz Republic described Kani TTS, an open-source speech model with 4 million parameters that runs on a standard laptop, was trained in six hours on eight GPUs and already supports more than 15 languages including Kyrgyz and Arabic . He said this proved world-class AI can be built in places such as Bishkek and announced SNAP, a Small Nations AI Powerhouses alliance to pool data, infrastructure and expertise .

The participant argues that language visibility in AI is a matter of future inclusion. Open-source language technologies can help ensure smaller languages are represented rather than erased from digital systems.

The speaker said that if a language is absent from the data, it is absent from the future . He then presented the open-source Kani TTS model, noting that any team can adapt it for their own languages and that it already supports numerous languages, providing a practical route to making smaller languages visible in AI .

The participant argues that environmental effects of AI infrastructure are localised and therefore cannot be managed solely at national or global level. Local governments are the ones that must deal with these impacts in practice.

The Global Cities Hub representative said that several ministers had mentioned the importance of data centres, but these centres also have significant environmental impacts . He stressed that those impacts are local and that local governments are the ones who must face them .

The participant argues that AI governance often talks about protecting young people without genuinely including them in decision-making. She says the real issue is concentrated corporate power and institutional dependence on large technology firms, not simply young people's access to platforms.

The youth participant asked those aged 30 or younger to stand and observed that there were very few young people present despite repeated claims that policy is being made for them . She argued that governance spaces focus on age verification and restrictions, while failing to understand how young people experience AI, and said public institutions are not bold enough to govern concentrated private power while continuing to sign large contracts with the same companies they are meant to regulate .

The participant argues that AI governance structures overlook local authorities even though deployment happens where people live. He presents local governments as essential actors because they generate and manage data, confront environmental burdens and implement rights in practice.

The participant from the Global Cities Hub said local governments are not part of the discussion even though AI is used and implemented at local level . He gave examples that cities produce much of the data used by AI, that local governments face environmental impacts from data centres and that human rights issues are also very important at local level .

Libya, speaking from an African perspective, argues that the AI divide is about far more than access to technology. It includes gaps in jobs, influence, infrastructure and data control, and therefore African states must be included in rule-making and benefit-sharing.

The speaker said Africa has 1.4 billion people but fewer than 2% of global data centres and even fewer AI patents . He argued these figures show inequitable distribution of resources and possibilities and called for effective African participation in designing the rules and principles that will govern the future .

Libya argues that the environmental costs of AI infrastructure must be recognised as part of global fairness. It points out that data centres consume large amounts of energy and create emissions, while African countries contribute little to emissions yet do not equally benefit from the infrastructure.

The speaker said data centres require a great deal of energy and electricity and lead to greenhouse gas emissions . He added that African countries are among those contributing least to greenhouse gas emissions and yet are not duly benefiting from these data centres .

Speaking for the G77 and China, Uruguay argues that AI governance must be development-oriented and must address structural inequalities directly. Without infrastructure, finance, skills and room to shape national policy, developing countries risk being marginalised further.

The statement said the scientific panel confirmed concerns that AI capabilities, infrastructure, data resources and expertise remain highly concentrated and that without sustained investment in infrastructure, skills, institutions, data and governance capacity, AI could reinforce existing inequalities . It also said developing countries need infrastructure, capacity, financing, technology and policy space to participate meaningfully in AI ecosystems .

The G77 and China argue that the foundational UN architecture for AI governance has now been established and attention should turn to implementation. They want more coherent UN action and practical mechanisms that connect discussion to delivery.

The statement said member states had already laid the foundation of an emerging UN architecture for AI governance through the General Assembly resolution and Global Digital Compact . It then said the next phase must focus on implementation, practical outcomes, strengthening capacity-building, innovative financing, making the global dialogue a platform that connects discussion with implementation, and strengthening coherence across the UN system .

The G77 and China present AI as a potentially important accelerator of sustainable development if governance is inclusive and supportive. They see its value across public services and productive sectors but warn that benefits will not arise automatically.

The statement said AI can accelerate progress towards the Sustainable Development Goals . It then listed improvements in public services, education, health systems, agricultural productivity, climate resilience, public administration, innovation and economic diversification as concrete benefits .

The Group of Friends argues that global AI governance must treat all countries as entitled to shape and use AI according to their own conditions. It also says support should be prioritised for developing countries and people in vulnerable situations.

The representative said all countries have the equal right to design, develop, deploy and use AI and to choose development paths suited to their national conditions, while safeguarding technological sovereignty, national security and development interests . She also said fairness, inclusiveness and non-discrimination should guide the dialogue and that priority should be given to developing countries, countries in special situations and people in vulnerable situations .

The Group of Friends supports a cooperative global framework that combines fairness with practical interoperability and resource sharing. It sees the UN as central to coordinating this international cooperation.

The representative said the United Nations should play the central coordinating role in international development cooperation . She also called for a global dialogue that maximises digital inclusion, enhances the interoperability of AI governance frameworks, standards and norms, and supports sharing and mutual benefit of AI resources through broad forms of cooperation .

The Group of Friends argues that inclusive governance must protect cultural and linguistic diversity. It treats multilingual AI as a concrete priority rather than an optional feature.

The representative emphasised the importance of promoting and preserving linguistic and cultural diversity and said the group supports the development of multilingual AI .

The Group of Friends sees openness in software, models and data access as a key way to broaden participation in AI. At the same time, it argues this should be balanced with respect for intellectual property rights.

The representative encouraged the development of open source software, open models and the sharing and mutual benefit of AI resources . She added that there should be non-discriminatory access to data, especially learning data, while respecting intellectual property rights .

Saudi Arabia argues that current concentration of AI compute and intellectual property is incompatible with an inclusive global future. It presents itself as willing to invest and cooperate to reduce this gap across the international system.

The minister said that 90% of compute and AI-related intellectual property are held in only two countries and asked whether that was an acceptable future for more than 190 UN member states . He then said Saudi Arabia had instructions from its leadership to be a main partner in reducing this gap and had made considerable investments to expand compute capacity over the coming years .

Costa Rica argues that mere consumption of imported AI systems does not amount to meaningful inclusion. Real inclusion requires the ability to adapt systems locally, govern data, shape standards and retain economic and public value.

The minister said some countries design AI, set standards and capture value while others barely access imported systems, and that this is not an inclusive digital future . She added that the challenge is to build decision-making capacity, adopt models in local languages and institutions, govern strategic data, influence international standards and retain a fair share of the value AI systems generate .

Costa Rica argues that AI should be judged not by technical sophistication but by whether it solves pressing human problems. It highlights humanitarian and security-related use cases as tests of AI’s real social value.

The minister said AI will not be justified by algorithmic sophistication but by its impact on people . She then said AI must help anticipate forced displacement, protect children, identify and respond to human crises and ensure sufficient infrastructure in the face of cyber war .

Costa Rica links AI governance to resilience against conflict and the protection of vulnerable groups. It sees cyber preparedness and child protection as important benchmarks for whether governance is truly human-centred.

The minister said AI must help protect children and ensure sufficient infrastructure in the face of cyber war .

Kenya argues that moving from principles to implementation is essential if AI is to expand opportunity rather than deepen existing divides. It focuses on the concrete ingredients countries need to participate meaningfully and govern AI safely.

The cabinet secretary said Kenya's approach is practical and that AI must expand opportunity and prosperity for all countries without deepening existing divides . He then listed affordable compute, open and representative datasets, safe and accessible models, stronger skills, meaningful participation in standard-setting, fair financing for infrastructure and research, and stronger public institutions as necessary priorities .

Kenya argues that the UN dialogue should not stand alone but should connect with other governance efforts at different levels. It therefore supports a more integrated ecosystem of AI discussions and implementation processes.

The cabinet secretary said this global dialogue should serve as a culmination of national, regional and international conversations on AI . He explicitly mentioned the Internet Governance Forum, the AI Action Summit, standards dialogues, the African Union and other regional processes as part of this wider ecosystem .

Kenya argues that AI inclusion requires systems to reflect African realities rather than forcing countries to consume externally designed tools. Without that, Global South countries will remain dependent and exposed to harms they did not create.

The cabinet secretary said the divide is felt in compute, data and language and that many AI systems do not reflect African languages, data or context . He warned that if this continues, countries in the Global South will remain consumers of technologies they did not shape while bearing risks and harms they did not create .

Botswana argues that developing countries cannot fully harness AI without adequate infrastructure and skills. It therefore wants the dialogue to deliver practical support and actionable recommendations rather than remain only conceptual.

The minister said Botswana and many developing countries face challenges in fully harnessing AI . He identified lack of adequate digital infrastructure and digital skills as a major constraint on effective adoption and use , and said the dialogue should create space for substantive exchange and practical solutions responding to developing countries' priorities .

Botswana presents AI as an important resource for national development across major service sectors. It links this to sustainable development and the ability to advance domestic priorities.

The minister said AI presents significant opportunities for advancing the Sustainable Development Goals . He added that Botswana recognises AI's capabilities and can strategically harness it to advance national development priorities across education, agriculture, healthcare and other key sectors .

Tanzania argues that developing countries must be included as equal participants in shaping global AI rules. It links this to practical support in skills, research, technology transfer and digital infrastructure.

The minister said developing countries must have an equal voice in shaping global AI policies, standards and governance frameworks . She then called for enhanced international cooperation in capacity building, technology transfer, research partnerships, innovation ecosystems and investment in digital infrastructure .

Tanzania presents AI as central to its long-term national development vision. It sees responsible AI as a driver of transformation across major public and economic sectors.

The minister said Tanzania views AI as a strategic enabler for Vision 2050 and the national digital economy framework . She listed agriculture, healthcare, education, financial services and public administration as key sectors where AI presents significant opportunities .

Uzbekistan argues that inclusive AI development depends on bringing developing countries into the agenda as participants. It sees investment in human capital as the main mechanism for ensuring broader benefit-sharing.

The minister said the future of AI governance should focus on inclusiveness and that all countries, including developing countries, should take part in the AI development agenda . He also said human capital is the most important capital Uzbekistan has and described heavy investment in education .

Uzbekistan presents AI as a practical tool for human development rather than only economic competition. It connects AI deployment to anti-poverty efforts, education and healthcare, supported by large-scale skills programmes.

The minister said AI should serve the global goals of reducing poverty, ending hunger and improving education and healthcare . He added that Uzbekistan had implemented a '5 million AI leaders' programme to upskill the population in using AI effectively for future achievements .

Uzbekistan sees open-source AI as a vehicle for broad social benefit, particularly in public-interest sectors. It argues that shared models can support global improvements in education and health.

The minister said open source is important and that open source models could be shared to help humanity improve education and healthcare .

Lebanon argues that sovereignty in AI means having the capability to understand, govern and secure AI, not isolation from the global system. It is therefore investing in skills, digital foundations, governance readiness and links with its diaspora.

The minister described NUMU, a national digital and AI capacity-building programme, as a human-centric effort to expand access to digital and AI skills . He also said Lebanon had revised the school curriculum to include responsible AI, was building digital state foundations, and was focusing on governance readiness, cyber readiness and data readiness so it could shape AI in line with its values .

Portugal argues that the governance challenge is not just regulation from outside but responsible implementation by governments themselves. It also sees effective international coordination between major institutions as necessary for coherent governance.

The minister said the real challenge is whether governments can deploy AI responsibly and at scale to improve people's lives . He later stated that governments should not simply regulate AI but become responsible adopters of AI, and that effective coordination between the UN, OECD, EU and other organisations remains essential .

Portugal presents AI as a tool for state modernisation and better public service delivery. It emphasises implementation in concrete administrative functions and policy experimentation through digital twins.

The minister said Portugal had approved a national AI agenda as an implementation agenda rather than another strategy document . He then listed uses in simplifying licensing procedures, modernising public procurement, improving healthcare, strengthening crisis response and building digital twins of public administration so governments can test policies before implementing them .

Portugal argues that language is part of digital sovereignty and that AI should reflect national and cultural realities. It therefore supports open language models built for specific linguistic communities.

The minister said 'language is also sovereignty' and announced Amalia, the first open LLM built specifically for Portuguese culture . He explained that it is designed for government, citizens, researchers and companies to build AI solutions in Portuguese, a language spoken by almost 300 million people .

France argues that inclusive AI governance must ensure AI reflects linguistic and cultural plurality. It treats multilingualism as one of its five central policy priorities.

The minister laid out five priorities and identified multilingualism as the second one, saying it was needed to make sure AI reflects the linguistic and cultural diversity of the world .

France presents open-source AI as a means of broadening access and reducing concentration. It links openness to the democratisation of this new technological capability.

In its five-priority framework, the minister identified open source models as the fifth priority and said they are needed to democratise access to this new energy .

France argues that responsible AI governance must confront both democratic harms and environmental limits. It identifies disinformation and sustainability as explicit policy priorities.

The minister listed fighting manipulation of information to preserve the integrity of democratic debate as the third priority . She also said the fourth priority is respecting the environment and added that countries facing energy problems must ensure AI deployment does not worsen the environmental situation .

Ambassador Tammsaar frames the governmental plenary as a space for states to exchange concrete experiences and policy approaches. He specifically links the session to inclusive and interoperable AI governance and to expectations for international cooperation.

He said the high-level governmental plenary segment provides an opportunity for the UN and governments to share national and regional experiences, priorities, best practices and approaches to AI governance, while reflecting on expectations for international cooperation . He then opened the segment as one focused on 'harnessing the benefits of AI for all through inclusive and interoperable approaches' .

Ambassador Tammsaar explains that the meeting is designed in stages so that broad stakeholder input feeds into a more structured multi-stakeholder discussion. This framing is meant to connect diverse viewpoints rather than isolate them.

He said the meeting would move from the governmental session into a high-level multi-stakeholder plenary segment . At the end, he thanked participants for their interventions and said the discussion would now move to a high-level moderated round table featuring representatives from governments, industry, civil society, academia, the technical community and international organisations .

Ambassador López emphasises procedural fairness and equal speaking opportunities in the dialogue. Her interventions focus on the practical rules needed to let as many participants as possible contribute.

She explained that speaking time would be limited to three minutes for everyone and that speakers were encouraged to respect the allotted time so that the largest possible number could be accommodated . She added that microphones would be switched off once time expired in order to ensure fairness and broad participation .

Ambassador López responds directly to criticism about inclusion by arguing that the process has already sought wide participation and should become even broader over time. She presents this first dialogue as the start of an ongoing and expanding process.

Responding to a participant, she said the dialogue had tried throughout the process to be as inclusive as possible and that no one had been excluded, with opportunities to participate both in person and online . She added that this was only the first dialogue and that she hoped even more stakeholders would be able to engage by the second one .