The transcript captures the closing ceremony of the first UN Global Dialogue on AI Governance, during which co-chairs Rein Tammsaar and Egriselda López reflected on the outcomes and significance of the event.
Tammsaar opened by highlighting the scale of participation, noting that more than 4,200 registered participants from 1,800 entities representing nearly 170 member states took part . He emphasised that the dialogue began with evidence from an Independent International Scientific Panel, whose report warned that AI capabilities are advancing faster than the ability to govern them, and that many countries lack the capacity to test, audit, or oversee these systems . The panel also affirmed that the benefits of AI are real in areas such as health, science, and food security, while the harms are already affecting real people, including children .
A key finding across the dialogue's four clusters was that the primary gap is not one of ambition or principles, but of practical mechanisms . Tammsaar stressed that only a fraction of the world's 7,000 languages have the foundations for meaningful inclusion in AI, and that environmental sustainability must be treated as a co-governance issue rather than a side topic . He confirmed that a second global dialogue is planned for 2027, with the expectation that participants will arrive with concrete, result-oriented commitments rather than restated principles .
López built on these remarks by stressing that the dialogue's purpose was not to replace existing AI governance efforts but to connect them and strengthen cooperation across the broader ecosystem . She recalled a statement from a representative of a small country who declared they were present "not to be a footnote," underscoring the dialogue's commitment to ensuring every stakeholder has the opportunity to shape the collective future .
Both co-chairs concluded by expressing gratitude to participants, the Joint Secretariat, the scientific panel, and their own teams, affirming that what had been built was not a conclusion but a foundation for ongoing international cooperation on AI governance .
Major Discussion Points
- The establishment of a new global platform for AI governance cooperation: The closing session emphasised that this dialogue represented the creation of a significant new multilateral forum, bringing together over 4,200 participants from nearly 170 member states, spanning governments, civil society, academia, and the private sector. The co-chairs stressed that no single country or institution can govern AI alone, making multilateralism essential. - The gap between AI capabilities and governance mechanisms: A recurring theme was the finding from the Independent International Scientific Panel that AI capabilities are advancing faster than the ability to verify and govern them, with many countries lacking the capacity to test, audit, or oversee these systems. The dialogue identified this as a gap not of ambition or principles, but of practical mechanisms. - Inclusivity, the AI divide, and capacity building: Discussions highlighted that the AI divide extends beyond mere access to technology, encompassing the capacity to shape, govern, and benefit from it. Specific concerns were raised about the fact that only a fraction of the world's 7,000 languages have the foundations for meaningful inclusion in AI, and that capacity building must mean co-creation rather than consumption. A representative of a small country was notably quoted as stating, "we are here not to be a footnote." - Protection of children and human rights in AI governance: The dialogue consistently returned to the need to protect children by design rather than as an afterthought, recognising them as the fastest adopters of the technology. Human rights discussions set priorities including independent justice systems to keep AI accountable and the protection of children's rights. - The path forward and commitment to continued cooperation: Both co-chairs underscored that this dialogue is a beginning, not an end, with a second UN Global Dialogue on AI Governance planned for May 2027. The challenge issued was to arrive at the next dialogue not with restated principles, but with named, supported, and result-oriented commitments. The growing ecosystem of initiatives, institutions, and communities was acknowledged, with the dialogue's role being to connect rather than replace these efforts. ---
Overall Purpose or Goal
- The discussion served as the closing ceremony of the first UN Global Dialogue on AI Governance, held in Geneva. Its primary purpose was to reflect on the outcomes of the two-day event, acknowledge the breadth and depth of participation, and articulate a shared vision for ongoing international cooperation on AI governance. The dialogue aimed to establish a multilateral platform where diverse stakeholders could identify common ground, address the practical gaps in AI governance, and lay the groundwork for future commitments ahead of the 2027 follow-up dialogue.
- --
Overall Tone
- The overall tone of the discussion was hopeful, collaborative, and forward-looking, with an undercurrent of sober realism. Both co-chairs, Rein Tammsaar and Egriselda López, struck a measured but optimistic note, acknowledging the genuine challenges posed by rapidly advancing AI capabilities whilst celebrating the achievement of convening such a broad and diverse global forum . The tone remained consistently constructive throughout, never becoming adversarial or pessimistic. Towards the end of both speeches, the register became more personal and heartfelt, with expressions of gratitude to teams, families, and participants , lending the closing moments a warm and sincere quality. The phrase "proof that meaningful cooperation on AI governance is not only necessary, but possible" encapsulates the prevailing spirit - cautiously optimistic and firmly committed to collective action.
Closing Ceremony of the First UN Global Dialogue on AI Governance
#
Opening and Scale of Participation
The closing ceremony of the first UN Global Dialogue on AI Governance brought together senior figures from across the United Nations system, with co-chair Egriselda López inviting to the podium Guy Ryder, Under-Secretary-General for Policy; Tomas Lamanauskas, Deputy Secretary-General of ITU; Mariya Gabriel, Assistant Director-General of UNESCO; and Amandeep Singh Gill, Under-Secretary-General and Special Envoy for Digital and Emerging Technologies .
The session was opened by co-chair Rein Tammsaar, who framed the occasion not merely as the conclusion of a meeting but as the establishment of a significant new multilateral platform for global cooperation on one of the defining issues of the age . He underscored the unprecedented scale of the event, noting that more than 4,200 registered participants from 1,800 entities representing nearly 170 member states had taken part, spanning heads of state and government, ministers, international organisations, civil society, academia, the technical community, and the private sector . Crucially, Tammsaar emphasised that what distinguished the dialogue was not only the breadth of participation but its depth .
#
Evidence Base and the Role of the Independent International Scientific Panel
A central feature of the dialogue was its grounding in scientific evidence. Tammsaar noted that proceedings began with the first report of the Independent International Scientific Panel, representing the minimum consensus of 40 leading international experts who answered only to the evidence . The panel's findings were described as sobering: AI capabilities are advancing faster than the ability to verify and govern them, and many countries still lack the capacity to test, audit, or oversee these systems . At the same time, the panel affirmed that the benefits of AI are real - in health, science, and food security - whilst making clear that the harms are not hypothetical but are already affecting real people, including children . Significantly, the panel urged participants not to wait for certainty before acting , a call that Tammsaar indicated had been heard .
Tammsaar also noted that a "dialogue of dialogues" session had just concluded, drawing together the many conversations held across the event and reflecting on convergence, complementarity, and mutual learning across the dialogue as a whole. Co-chair López later reinforced the theme of scientific grounding, expressing appreciation for the panel's engagement throughout the dialogue and affirming that "sound governance must be informed by sound science," with the panel's contribution expected to remain invaluable as the process moves forward .
#
The Gap of Practical Mechanisms
One of the most analytically significant observations of the closing session came from Tammsaar's identification of the nature of the governance deficit. Across the dialogue's four clusters, a common thread had emerged: the gap facing the international community is not one of ambition or principles, but of practical mechanisms . He noted that in room after room, delegations had moved past obstruction towards implementation, with discussions covering national AI strategies and agencies, legal frameworks, safeguards for minors, teacher training, and public sector capacity . This framing - distinguishing between normative progress and operational readiness - set the intellectual tone for the closing session and implicitly challenged participants to focus their future efforts on concrete, deliverable tools rather than further declarations of intent.
#
The AI Divide, Inclusivity, and Linguistic Representation
The dialogue surfaced a broader and more complex understanding of the AI divide than is often acknowledged in international policy discussions. Tammsaar stressed that the divide is not only about access to technology, but about the capacity to shape, govern, and benefit from it . He argued that capacity building must mean co-creation rather than consumption - a distinction that challenges the dominant paradigm in which technology flows from developed to developing nations, with recipient countries positioned as passive beneficiaries. Particularly striking was his observation that only a fraction of the world's 7,000 languages have the foundations for meaningful inclusion in AI , highlighting a profound cultural and epistemic dimension of exclusion that extends well beyond infrastructure or economic access. Environmental sustainability was also identified as a co-governance concern rather than a peripheral topic , signalling that the dialogue had achieved a degree of consensus on integrating environmental considerations into the core of AI governance frameworks.
#
Protection of Children and Human Rights
The protection of children emerged as a recurring and foundational concern throughout the dialogue. Tammsaar noted that children, as the fastest adopters of AI technology, must be protected by design rather than as an afterthought . The morning's discussions on safe, secure, and trustworthy AI had returned to the scientific evidence base and its role in shaping safety frameworks through a shared minimum baseline and common technical standards . Human rights discussions had set priorities for the years ahead, including independent justice systems to keep AI accountable to humans and the protection of children's rights . These themes were woven throughout the dialogue's four clusters, reflecting a broad convergence among participants that child protection and human rights accountability are not niche concerns but foundational governance requirements.
#
Multilateralism as Essential, Not Optional
Tammsaar was unequivocal on the question of multilateral cooperation, stating plainly that no single country, company, or institution can govern AI alone, and that multilateralism in this domain is not optional . He concluded his remarks by affirming that the dialogue had provided proof that meaningful cooperation on AI governance is not only necessary but possible . He also invoked a measure of success that had been offered during the discussions and attributed to President Kari of Estonia: not who uses AI first, but who uses it wisely - a formulation that directly challenges the competitive framing that often dominates geopolitical discourse on AI and redefines success in terms of quality, ethics, and human benefit.
At the opening of the dialogue, Tammsaar had expressed the hope that it might spark a "San Francisco moment" for AI; in his closing remarks, he reflected on that hope, expressing confidence that the platform had been created and the momentum established, leaving it to participants to carry the work forward .
#
López on Connecting the Governance Ecosystem
Co-chair Egriselda López built on Tammsaar's remarks by offering a complementary framing of the dialogue's role and significance. She emphasised that the task of the dialogue is not to replace the growing ecosystem of different initiatives, institutions, and communities contributing to AI governance, but to connect them, build greater understanding, identify complementarities, and strengthen cooperation across this evolving landscape . She specifically highlighted the breadth of this ecosystem, encompassing national strategies, regional initiatives, international organisations establishing new bodies, technical communities, and other AI governance processes around the world. This positioning - as a connector and amplifier rather than a replacement or competitor - reflected a diplomatically astute acknowledgement of the complexity of the global AI governance landscape and was designed to sustain the buy-in of participants representing a wide range of existing processes and bodies . López stressed that the relationships and partnerships formed during the dialogue should continue to grow and deliver practical cooperation , and that the true measure of the first dialogue will not be only what was heard, but what participants choose to do once they leave .
#
Inclusive Participation and the Voice of Smaller Nations
One of the most resonant moments of López's closing remarks came when she recalled a statement made during the governmental segment by a representative of a small country: "we are here not to be a footnote" . She noted that this thought had stayed with her, because it captured the fundamental purpose of the dialogue - ensuring that every country, every institution, and every stakeholder has the opportunity not only to be heard, but to shape the collective future . This observation gave concrete human expression to the abstract principle of equity and implicitly set a standard against which the success of future dialogues should be measured: not merely whether all countries attended, but whether all countries genuinely influenced the outcomes. López framed the responsibility now shared by all participants as one of carrying the work forward from their own governments, institutions, organisations, and communities .
#
The Path Forward: A Foundation, Not a Conclusion
Both co-chairs were emphatic that the dialogue represents a beginning rather than an end. Tammsaar stated directly that "this is not the end of the process. It is the beginning of one" , pointing to the continuation of the global dialogue planned for 2027 . He issued a clear challenge: participants should arrive at the next dialogue not with restated principles but with named, supported, and result-oriented commitments . He also noted that the work does not pause in the interim, and that the richness of the discussions provides substantial material for the co-chairs' summary, which must faithfully reflect both the convergence and the diversity of views expressed . López echoed this framing directly, stating that "what we have built here is not a conclusion. It is a foundation" , and expressing hope that the spirit of openness, mutual respect, and shared purpose that characterised the discussions will continue to inspire cooperation well beyond Geneva . She confirmed that the second UN Global Dialogue on AI Governance is scheduled for May 2027 .
#
Expressions of Gratitude and Personal Reflections
Both co-chairs concluded with sincere expressions of gratitude to the many individuals and institutions that made the dialogue possible. Tammsaar thanked Secretary-General Guterres and President Baerbock for their leadership and trust, his fellow co-chair Ambassador López, the Joint Secretariat co-coordinated by ITU and UNESCO, and every participant who gave their time, expertise, and conviction to the process . He offered special thanks to team members Greta, Robert, and Walter, who worked for 249 days - and many nights - to make the dialogue happen .
López extended her appreciation to every participant, both in person and online, and to those who travelled to Geneva specifically for the occasion, as well as to leaders from government, industry, civil society, academia, the technical community, and international organisations, and to every speaker, moderator, and colleague working behind the scenes . She offered particular thanks to the members of the Independent International Scientific Panel and to the Joint Secretariat, ITU, UNESCO, and the Office of the Secretary-General . She also paid personal tribute to the families of both co-chairs, who had walked the path alongside them through endless days, long nights, and constant travelling. She paid further tribute to the teams of Estonia and El Salvador - two small missions that she described as having shown enormous responsibility, professionalism, generosity, and exceptional commitment to the cause . In a personal reflection, López described it as a true privilege to have co-chaired the process, noting that it had brought her into contact with remarkable people from across the world, united by a genuine commitment to putting AI at the service of humanity, and that this left her with a profound sense of hope . Both co-chairs closed by wishing participants a safe journey home and expressing their anticipation of meeting again in 2027 .
#
Overall Significance
Taken together, the closing remarks of the first UN Global Dialogue on AI Governance conveyed a tone that was hopeful and forward-looking whilst remaining soberly realistic about the challenges ahead. The dialogue had demonstrated that approaches to AI governance may differ across nations and stakeholders, but that there is a shared recognition of the need for international cooperation and of the value of a common space where differences can be expressed constructively . The co-chairs left participants with a clear message: the platform has been created, the momentum established, and the proof of concept delivered - that meaningful cooperation on AI governance is not only necessary, but possible . The work of translating that possibility into named, supported, and result-oriented commitments now falls to every government, institution, and community represented in the room .
Broad and inclusive participation
Arg. 1The dialogue brought together over 4,200 registered participants from nearly 170 member states, representing governments, civil society, academia, and the private sector, demonstrating unprecedented breadth and depth of participation. This scale of engagement underscored the global significance of the event as a new platform for international cooperation on AI governance.
More than 4,200 registered participants from 1,800 entities representing nearly 170 member states attended, including heads of state and government, ministers, international organisations, civil society, academia, the technical community, and the private sector . Tammsaar emphasised that what stood out was not only the breadth of participation but also its depth .
AI capabilities outpacing governance
Arg. 2The Independent International Scientific Panel's first report, representing the minimum consensus of 40 leading international experts, found that AI capabilities are advancing faster than our ability to verify and govern them. Many countries still lack the capacity to test, audit, or oversee these systems, making urgent action necessary.
The panel's report delivered a sobering message: capabilities are advancing faster than our ability to verify and govern them, and many countries still lack the capacity to test, audit, or oversee these systems . The panel asked participants not to wait for certainty before acting .
on: Sound AI governance must be grounded in scientific evidence
Gap of practical mechanisms
Arg. 3The gap in AI governance is not one of ambition or principles, but of practical mechanisms for implementation. Across the dialogue's four clusters, delegations moved past obstruction towards concrete implementation measures including national AI strategies, legal frameworks, and public sector capacity.
Tammsaar noted that across four clusters, a common thread emerged: the gap faced is not a gap of ambition or principles but of practical mechanisms . Delegations moved towards implementation including national AI strategies and agencies, legal frameworks, safeguards for minors, teacher training, and public sector capacity .
on: Framing of the dialogue's primary role
Capacity and inclusion gap
Arg. 4The AI divide extends beyond mere access to technology; it encompasses the capacity to shape, govern, and benefit from AI. Only a fraction of the world's 7,000 languages have the foundations for meaningful inclusion in AI, highlighting deep structural inequalities.
Tammsaar stated that the AI divide is not only about access to technology but about the capacity to shape, govern, and benefit from it, and that capacity building must mean co-creation, not consumption . He further noted that only a fraction of the world's 7,000 languages have the foundations for meaningful inclusion in AI .
Children's protection by design
Arg. 5Children, as the fastest adopters of AI technology, must be protected by design rather than as an afterthought. Independent justice systems must keep AI accountable to humans, and the protection of children's rights must be a central priority in AI governance frameworks.
Tammsaar emphasised that children, as the fastest adapters of this technology, must be protected by design, not taken as an afterthought . He also noted that conversations on human rights set priorities for the years ahead, including independent justice systems that keep AI accountable to humans and the protection of children's rights .
Beginning of a process, not an end
Arg. 6The dialogue marks the beginning of a process rather than its conclusion, with a continuation planned for 2027. Participants are challenged to arrive at the next dialogue not with restated principles but with named, supported, and result-oriented commitments.
Tammsaar stated clearly that this is not the end of the process but the beginning of one, with a continuation of the global dialogue planned for 2027 . He noted that the richness of the discussions provides substantial material for the co-chairs' summary, which must faithfully reflect both convergence and diversity .
on: The cooperative spirit and relationships built must continue beyond Geneva
Multilateralism as essential
Arg. 7No single country, company, or institution can govern AI alone, making multilateralism not optional but essential. The dialogue has demonstrated that meaningful cooperation on AI governance is not only necessary but possible.
Tammsaar asserted that no single country, company, or institution can govern this technology alone, and that multilateralism is not optional . He concluded that the dialogue provides proof that meaningful cooperation on AI governance is not only necessary but possible .
on: Multilateralism and inclusive international cooperation are essential for AI governance
Connecting the AI governance ecosystem
Arg. 1The dialogue served as an important milestone in connecting a growing ecosystem of different initiatives, institutions, and communities contributing to AI governance, rather than replacing them. Its task is to build greater understanding, identify complementarities, and strengthen cooperation across this evolving ecosystem.
López emphasised that there is a growing ecosystem of different initiatives, institutions, and communities that each contribute valuable perspectives on AI governance . She stated that the task of the dialogue is not to replace these valuable efforts but to connect them and build greater understanding, identify complementarities, and strengthen cooperation .
on: Multilateralism and inclusive international cooperation are essential for AI governance
on: Framing of the dialogue's primary role
Science-informed governance
Arg. 2Sound governance must be informed by sound science, and the Independent International Scientific Panel's expertise and engagement throughout the dialogue enriched discussions. The panel's contribution will remain invaluable as the AI governance process moves forward.
López expressed appreciation to the members of the Independent International Scientific Panel on AI, noting that their expertise, commitment, and willingness to engage throughout the dialogue enriched discussions . She stated that sound governance must be informed by sound science and that the panel's contribution will continue to be invaluable as the process moves forward .
on: Sound AI governance must be grounded in scientific evidence
Inclusive shaping of the future
Arg. 3Every country, institution, and stakeholder must have the opportunity not only to be heard but to shape the collective future. This principle was powerfully illustrated by a representative of a small country who stated 'we are here not to be a footnote', underscoring the importance of inclusive participation in AI governance.
López recounted hearing the representative of a small country say during the governmental segment, 'we are here not to be a footnote', a thought that stayed with her . She emphasised that the dialogue has been about ensuring that every country, every institution, and every stakeholder has the opportunity not only to be heard but to shape the collective future .
on: Multilateralism and inclusive international cooperation are essential for AI governance
Measuring success through action
Arg. 4The true measure of the first dialogue will not be what was heard but what participants choose to do once they leave. The relationships and partnerships formed during the dialogue need to continue delivering practical cooperation beyond Geneva.
López stated that the true measure of the first dialogue will not be only what was heard but what participants choose to do once they leave . She noted that the relationships built and partnerships formed should continue to grow and deliver practical cooperation .
on: Future commitments must be concrete and result-oriented, not merely principled
Sustaining cooperative spirit
Arg. 5The spirit of openness, mutual respect, and shared purpose that characterised the discussions should continue to inspire cooperation well beyond Geneva. National strategies, regional initiatives, and international organisations all have a role to play in sustaining this cooperative momentum.
López expressed hope that the spirit of openness, mutual respect, and shared purpose that characterised the discussions will continue to inspire cooperation well beyond Geneva . She referenced the broader ecosystem of national strategies, regional initiatives, international organisations starting up bodies, technical communities, and other AI governance processes around the world as part of this ongoing effort .
on: The cooperative spirit and relationships built must continue beyond Geneva
Session Knowledge Graph
Speakers · Topics · Arguments · Relationships
Both co-chairs explicitly framed the first UN Global Dialogue on AI Governance as the start of an ongoing process rather than a final outcome. Tammsaar stated 'This is not the end of the process. It is the beginning of one' , pointing to the continuation planned for 2027 . López echoed this directly: 'What we have built here is not a conclusion. It is a foundation' , and emphasised that the true measure of the dialogue will be what participants choose to do once they leave .
Beginning of a process, not an end
Measuring success through action
Both speakers affirmed the centrality of the Independent International Scientific Panel's work to the governance process. Tammsaar highlighted the panel's sobering finding that capabilities are advancing faster than our ability to verify and govern them , and noted the panel asked participants not to wait for certainty before acting . López reinforced this by expressing appreciation for the panel's contribution and stating that 'sound governance must be informed by sound science' , emphasising the panel's ongoing invaluable role .
AI capabilities outpacing governance
Science-informed governance
Both co-chairs stressed that the next phase of work must move beyond restating principles towards tangible outcomes. Tammsaar issued a challenge to arrive at the 2027 dialogue 'not with restated principles but with named, supported and result-oriented commitments' . López similarly noted that the true measure of the first dialogue will not be only what was heard but 'what we choose to do once we leave it' , and that partnerships formed should 'continue to deliver practical cooperation' .
Beginning of a process, not an end
Measuring success through action
Both speakers converged strongly on the necessity of multilateral, inclusive cooperation. Tammsaar asserted that 'no single country, company or institution can govern this technology alone' and that 'multilateralism here is not optional' , concluding that the dialogue proved meaningful cooperation is 'not only necessary, but possible' . López emphasised that the dialogue's task is to connect the growing ecosystem of initiatives rather than replace them , and that every country, institution, and stakeholder must have the opportunity not only to be heard but to shape the collective future .
Multilateralism as essential
Connecting the AI governance ecosystem
Inclusive shaping of the future
Both co-chairs called for the momentum and relationships generated by the dialogue to be sustained after the event. Tammsaar noted that 'the work does not pause' and that the richness of discussions provides substantial material for the co-chairs' summary . López expressed hope that 'the spirit of openness, mutual respect, and shared purpose that has characterised these discussions will continue to inspire cooperation well beyond Geneva' , and that relationships and partnerships formed should continue to grow and deliver practical cooperation .
Beginning of a process, not an end
Sustaining cooperative spirit
Both speakers shared a strong concern for ensuring that all countries, particularly smaller and less-resourced ones, are not merely passive recipients but shapers of AI governance. Tammsaar noted that the AI divide is not only about access to technology but about 'the capacity to shape, govern, and benefit from it', and that capacity building must mean 'co-creation, not consumption' . López powerfully illustrated this with the words of a small country's representative: 'we are here not to be a footnote' , affirming that the dialogue has been about ensuring every country and stakeholder has the opportunity to shape the collective future . Both speakers highlighted the significance of the breadth and diversity of participation as a key achievement of the dialogue. Tammsaar emphasised that what stood out was not only the breadth but the depth of participation, with over 4,200 registered participants from nearly 170 member states . López built on this by recognising the growing ecosystem of different initiatives, institutions, and communities contributing to AI governance, and framing the dialogue's role as connecting rather than replacing these efforts . Both speakers identified the translation of dialogue into practical implementation as the central challenge going forward. Tammsaar observed that the gap in AI governance is 'not a gap of ambition or of principles' but 'a gap of practical mechanisms' , noting that delegations moved towards concrete implementation measures . López reinforced this by stating that the true measure of the first dialogue will not be only what was heard but what participants choose to do once they leave , and that the responsibility is to carry the work forward from their own governments, institutions, and communities .
While environmental sustainability might not be expected as a prominent theme in a closing ceremony focused on AI governance achievements, Tammsaar explicitly noted that 'environmental sustainability is a co-governance, not a side topic' . This signals an unexpected degree of consensus within the dialogue that environmental concerns must be integrated into the core of AI governance frameworks rather than treated as peripheral, reflecting a broader convergence among participants that was notable enough to be highlighted in the closing remarks.
The strong and repeated emphasis on children's rights and protection in the context of AI governance represents a notable area of consensus that might not have been anticipated as a central theme of a high-level intergovernmental AI governance dialogue. Tammsaar highlighted that children, as 'the fastest adapters of this technology must be protected by design, not taken as an afterthought' , and that protection of children's rights was set as a priority for the years ahead . This framing of children as a primary governance concern, rather than a secondary welfare issue, reflects an unexpected convergence across the dialogue's participants.
The explicit recognition that only a fraction of the world's 7,000 languages have the foundations for meaningful inclusion in AI represents an unexpected area of consensus within the dialogue. Linguistic inclusion is rarely foregrounded as a core AI governance issue in high-level international settings, yet Tammsaar's inclusion of this point in the closing summary suggests it achieved sufficient prominence and agreement among participants to warrant mention alongside more conventional governance concerns such as legal frameworks and national strategies.
The closing session of the first UN Global Dialogue on AI Governance demonstrated a very high level of consensus between the two co-chairs, Rein Tammsaar and Egriselda López, across all major themes. Both speakers agreed that: the dialogue represents a beginning rather than a conclusion ; scientific evidence must underpin governance ; practical mechanisms and concrete commitments are needed beyond principles ; multilateralism is essential and non-optional ; and the cooperative spirit must be sustained beyond Geneva . Additional areas of convergence included the importance of inclusive participation, the AI capacity and inclusion gap, children's protection by design, environmental sustainability as a core governance concern, and linguistic inclusion as a structural challenge.
Tammsaar frames the dialogue's primary achievement as identifying and beginning to address the gap of practical mechanisms in AI governance, emphasising that delegations moved 'past obstruction towards implementation' and that the gap is 'not a gap of ambition or of principles' but 'of practical mechanisms' . López, by contrast, frames the dialogue's primary role as connecting an already-existing ecosystem of initiatives, stating that 'the task of this dialogue is not to replace these valuable efforts, but to connect them' . Tammsaar's framing positions the dialogue as a problem-solving forum addressing concrete governance deficits, while López's framing positions it as a coordination and networking platform within a broader landscape.
Gap of practical mechanisms
Connecting the AI governance ecosystem
In a closing ceremony where both speakers might be expected to be fully aligned, there is a subtle and unexpected difference in how prominently each speaker foregrounds the role of scientific evidence. Tammsaar devotes considerable attention to the Independent International Scientific Panel's findings, describing its report as delivering a 'sobering' message and noting that 'the panel asked us not to wait for certainty before acting' , positioning scientific evidence as a central driver of the dialogue's urgency. López, while acknowledging the panel and affirming that 'sound governance must be informed by sound science' , treats this as one element among many in her closing remarks , giving comparatively greater weight to the political and relational dimensions of the dialogue. This subtle divergence is unexpected given that both are co-chairs of the same process and might have been expected to present a fully unified narrative on the centrality of science.
The transcript represents closing remarks delivered by two co-chairs of the first UN Global Dialogue on AI Governance. Both speakers — Rein Tammsaar and Egriselda López — are broadly and substantially aligned in their assessments of the dialogue's achievements, its significance, and the path forward. The primary areas of subtle divergence relate to: (1) how the dialogue's role is framed (practical mechanism-building versus ecosystem-connecting); (2) the relative emphasis placed on scientific evidence versus political and relational dimensions of cooperation; and (3) the conceptualisation of inclusive participation (quantitative and structural versus qualitative and political). There are no explicit disagreements or contradictions between the speakers.
Both speakers agree that the dialogue represents a beginning rather than a conclusion and that future action is essential. Tammsaar states explicitly that 'this is not the end of the process. It is the beginning of one' and challenges participants to arrive at the 2027 dialogue 'not with restated principles but with named, supported and result-oriented commitments' . López similarly states that 'what we have built here is not a conclusion. It is a foundation' and that 'the true measure of this first dialogue will not be only what we listen, what we heard. It will be what we choose to do once we leave it' . However, they differ subtly in emphasis: Tammsaar focuses on the challenge of moving from principles to concrete commitments , while López focuses on the continuation of relationships and partnerships formed during the dialogue , suggesting a slightly different view of what 'action' means in practice.
Beginning of a process, not an end Measuring success through action
Both speakers agree on the necessity of international cooperation and multilateralism in AI governance. Tammsaar makes a strong structural argument that 'no single country, company or institution can govern this technology alone' and that 'multilateralism here is not optional' , grounding this in the evidence of the dialogue's achievements. López agrees but frames the cooperative spirit in more relational and cultural terms, expressing hope that 'the spirit of openness, mutual respect, and shared purpose that has characterised these discussions will continue to inspire cooperation well beyond Geneva' . Tammsaar's framing is more structural and institutional, while López's is more values-based and relational, reflecting a subtle difference in how they conceptualise the foundations of multilateral cooperation.
Multilateralism as essential Sustaining cooperative spirit
Both speakers emphasise the importance of inclusive participation, but with different focal points. Tammsaar highlights the quantitative breadth of participation — 'more than 4,200 registered participants from 1,800 entities representing nearly 170 member states' — and the structural capacity gap, noting that the AI divide is 'about the capacity to shape, govern, and benefit from' AI . López grounds inclusion in a more qualitative and political dimension, recounting the statement of a small country's representative — 'we are here not to be a footnote' — and emphasising that every country and stakeholder must have the opportunity 'not only to be heard, but to shape our collective future' . Tammsaar's approach is more data-driven and structural, while López's is more politically and humanistically oriented.
Broad and inclusive participation Inclusive shaping of the future
- The first UN Global Dialogue on AI Governance successfully convened over 4,200 registered participants from nearly 170 member states, representing an unprecedented breadth and depth of participation from governments, civil society, academia, the technical community, and the private sector.
- The Independent International Scientific Panel's first report, representing minimum consensus among 40 leading international experts, concluded that AI capabilities are advancing faster than our ability to verify and govern them, and that many countries lack the capacity to test, audit, or oversee AI systems.
- The primary gap in AI governance is not one of ambition or principles, but of practical mechanisms, with delegations moving towards implementation including national AI strategies, legal frameworks, safeguards for minors, and public sector capacity building.
- The AI divide extends beyond access to technology and encompasses the capacity to shape, govern, and benefit from AI, with only a fraction of the world's 7,000 languages having the foundations for meaningful inclusion.
- Children, as the fastest adopters of AI technology, must be protected by design rather than as an afterthought, and independent justice systems must keep AI accountable to humans.
- No single country, company, or institution can govern AI alone, making multilateralism not optional but essential; the dialogue demonstrated that meaningful international cooperation on AI governance is both necessary and achievable.
- The dialogue's role is not to replace existing AI governance initiatives, institutions, and communities, but to connect them, identify complementarities, and strengthen cooperation across the evolving ecosystem.
- Sound AI governance must be informed by sound science, with the Independent International Scientific Panel's contributions considered invaluable to the ongoing process.
- Every country, institution, and stakeholder must have the opportunity not only to be heard but to actively shape the collective future of AI governance, ensuring no nation becomes a mere footnote in the process.
- Environmental sustainability was identified as a co-governance issue, not a peripheral topic, within the broader AI governance framework.
- The true measure of success of this first dialogue will not be what was heard during the discussions, but what participants choose to do once they leave, with relationships and partnerships needing to deliver practical cooperation.
“The gap we face today is not a gap of ambition or of principles. It is a gap of practical mechanisms.”
“Capabilities are advancing faster than our ability to verify and govern them, and many countries still lack the capacity to test, audit, or oversee these systems.”
“That capacity building must mean co-creation, not consumption.”
“Only a fraction of the world's 7,000 languages have the foundations for meaningful inclusion in AI.”
“Children — the fastest adapters of this technology — must be protected by design, not taken as an afterthought.”
“Not who uses AI first, but who uses it wisely.”
“I heard the representative of a small country say, 'we are here not to be a footnote.' And that thought has stayed with me.”
“What we have built here is not a conclusion. It is a foundation.”
“The task of this dialogue is not to replace these valuable efforts, but to connect them and to build greater understanding, identify the complementarities, and strengthen cooperation across this evolving and rich ecosystem.”
How can the gap between rapidly advancing AI capabilities and the ability to verify and govern them be practically closed?
The Independent International Scientific Panel highlighted that capabilities are advancing faster than governance mechanisms can keep up with. This represents a critical structural challenge that requires concrete follow-up research and policy development to address before the next dialogue in 2027.
What practical mechanisms can be developed to allow countries, particularly those with limited capacity, to test, audit, and oversee AI systems?
Many countries currently lack the technical infrastructure to independently assess AI systems. Identifying and developing scalable, accessible oversight mechanisms is essential for equitable global AI governance and was flagged as an urgent gap.
How can AI capacity building be structured as co-creation rather than mere consumption, particularly for developing nations?
The dialogue surfaced a distinction between simply providing access to AI technology and enabling countries to actively shape and govern it. Further research is needed into models of genuine co-creation that empower all nations rather than reinforcing existing power imbalances.
What steps can be taken to ensure meaningful AI inclusion for speakers of the world's approximately 7,000 languages, the vast majority of which currently lack foundational AI support?
Only a fraction of the world's languages have the linguistic data and infrastructure needed for AI inclusion. This represents a significant equity gap requiring dedicated research into multilingual AI development and language resource creation.
How should environmental sustainability be integrated as a core co-governance concern within AI governance frameworks, rather than treated as a peripheral topic?
The dialogue noted that environmental sustainability must be embedded in AI governance rather than addressed as an afterthought. Further research is needed on the environmental footprint of AI systems and how governance frameworks can mandate sustainable practices.
How can AI systems be designed to protect children by default, given that children are among the fastest adopters of AI technology?
Children's rights and safety in AI contexts were identified as a priority area. Research is needed into design standards, regulatory frameworks, and safeguards that embed child protection at the architectural level of AI systems rather than as a retrospective measure.
What shared minimum baseline and common technical standards for AI safety can be established through the scientific evidence base?
The dialogue called for safety frameworks grounded in scientific evidence, but the specifics of what constitutes a shared minimum baseline remain to be defined. Further technical and policy research is needed to develop standards that can achieve broad international consensus.
How can independent justice systems be structured to keep AI accountable to humans and protect human rights, including children's rights?
The human rights discussions identified the need for independent judicial and accountability mechanisms for AI. Research into legal frameworks, institutional design, and enforcement mechanisms is required to make this a practical reality across different national contexts.
What named, supported, and result-oriented commitments can countries and institutions bring to the second UN Global Dialogue on AI Governance in 2027?
A direct challenge was issued to move beyond restated principles to concrete, measurable commitments by 2027. This requires follow-up work by all participating entities to translate dialogue outcomes into specific, accountable pledges with defined deliverables.
How can the growing ecosystem of AI governance initiatives, institutions, and communities be better connected to identify complementarities and strengthen cooperation?
The co-chair highlighted that the dialogue's role is not to replace existing efforts but to connect them. Further research and mapping of the global AI governance landscape is needed to identify overlaps, gaps, and opportunities for meaningful coordination.
How can the process ensure that smaller and less powerful countries are shapers of AI governance rather than mere footnotes in the process?
A representative of a small country expressed the desire not to be a footnote, which the co-chair highlighted as a central concern. Research into inclusive governance design and mechanisms for equitable participation is needed to ensure all nations have genuine influence.
How can sound science continue to inform AI governance as the Independent International Scientific Panel's work progresses?
The co-chair emphasised that sound governance must be informed by sound science. Ongoing research into how scientific findings can be systematically integrated into policy processes, and how the panel's work can remain relevant and authoritative, is a key area for further development.