AI Governance Dialogue: Steering the future of AI
10 Jul 2025 14:30h - 14:45h
AI Governance Dialogue: Steering the future of AI
Session at a glance
Summary
This discussion features Doreen Bogdan Martin, Secretary General of the ITU, delivering a keynote address on AI governance at the AI for Good Global Summit. Martin emphasizes that the world is experiencing a truly transformational moment in artificial intelligence, with the global AI race reshaping diplomacy, economics, and labor markets while exposing critical gaps in AI development and deployment capabilities. She highlights the positive potential of AI by citing the Nobel Prize-winning AlphaFold system, which modeled over 200 million proteins and made this data freely available to accelerate global drug discovery and medical research.
Martin argues that this transformative moment demands inclusive, forward-looking governance that drives innovation while building public trust and minimizing risks. She references the UN’s adoption of the Pact for the Future and Global Digital Compact as frameworks that serve as a compass for equitable AI development. However, she stresses that practical governance mechanisms are needed to actually steer AI progress toward shared benefits.
The discussion outlines three key elements for effective AI governance: inclusion, capacity, and standards. Martin notes that over 100 countries lack meaningful voice in global AI governance discussions, despite having 170 countries represented at the summit. She emphasizes the need to build capacity in developing countries through infrastructure access and skills development, highlighting the launch of the AI Skills Coalition. Additionally, she advocates for technical standards that translate high-level commitments into operational safeguards, noting recent consultations on AI testing and validation.
Martin concludes by emphasizing that AI governance is a shared multi-stakeholder responsibility requiring coordination between governments, industry, academia, civil society, and the UN system to navigate the AI future together under common principles.
Keypoints
**Major Discussion Points:**
– **AI Governance as a Transformational Moment**: The discussion emphasizes that we are at a critical juncture where AI governance must be steered in the right direction, using examples like AlphaFold’s protein modeling breakthrough to illustrate AI’s positive potential for scientific discovery and global benefit.
– **Inclusion in Global AI Governance**: A major concern is that over 100 countries lack meaningful participation in global AI governance discussions, despite having 170 countries represented at the summit. The speaker advocates for leveraging the UN’s global reach to make AI governance truly inclusive of all perspectives and contexts.
– **Capacity Building and Infrastructure**: The need to equip policymakers and public administrators, especially in developing countries, with skills to assess, procure, and deploy AI systems. This includes access to compute power, data centers, and technical infrastructure, as well as human capacity through initiatives like the AI Skills Coalition.
– **Technical Standards and Implementation**: Moving beyond high-level principles to create concrete technical standards that translate commitments into operational safeguards, emphasizing the need for multi-stakeholder collaboration in AI testing and validation methodologies.
– **Multi-stakeholder Responsibility**: The discussion outlines specific roles for different actors – governments (laws and rights protection), industry (responsible development), academia (evaluation and testing), civil society (advocacy), and the UN system (coordination and universal values).
**Overall Purpose:**
The discussion aims to advocate for comprehensive, inclusive AI governance that ensures the benefits of AI are shared globally while minimizing risks. The speaker is calling for coordinated international action to establish practical governance mechanisms that go beyond declarations to create real-world implementation frameworks.
**Overall Tone:**
The tone is inspirational and urgent, maintaining an optimistic yet realistic perspective throughout. The speaker uses maritime metaphors (compass, captain’s wheel, steering ships) to create a sense of collective navigation toward a shared destination. The tone remains consistently hopeful and collaborative, emphasizing partnership and shared responsibility rather than competition, and concludes with a unifying message about looking forward together.
Speakers
– Doreen Bogdan Martin – Secretary General of the ITU (International Telecommunication Union)
Additional speakers:
– His Excellency, Mr. Karis – President of Estonia
– His Excellency, Engineer Majed al-Mezmar – Director General of the Telecommunications and Digital Government Regulatory Authority of the UAE, Co-chair of AI Governance Dialogue
– Madame Anne Bouvereau – France’s Special Envoy for Artificial Intelligence, Co-chair of AI Governance Dialogue
– Fred – (Role/title not specified, appears to be involved in summit organization)
– Minister George – Minister from Ghana (full name not provided)
Full session report
# Report: AI Governance Keynote Address at the AI for Good Global Summit
## Executive Summary
This report summarizes a keynote address delivered by Doreen Bogdan Martin, Secretary General of the International Telecommunication Union (ITU), at the AI for Good Global Summit. Martin’s presentation focused on establishing comprehensive AI governance frameworks through three key pillars: inclusion, capacity building, and technical standards. She emphasized that the world is experiencing a transformational moment in artificial intelligence that requires coordinated international action to ensure AI benefits are shared globally while managing associated risks.
## Speaker and Context
The presentation was delivered by Doreen Bogdan Martin, Secretary General of the ITU, to representatives from 170 countries at the AI for Good Global Summit. Martin referenced contributions from various officials including His Excellency Mr. Karis (President of Estonia), His Excellency Engineer Majed al-Mezmar (Director General of the Telecommunications and Digital Government Regulatory Authority of the UAE), Madame Anne Bouvereau (France’s Special Envoy for Artificial Intelligence), and Minister George from Ghana.
## Key Arguments and Framework
### AI as a Transformational Moment
Martin established that the global AI race is fundamentally reshaping diplomacy, economics, and labor markets while exposing critical gaps in countries’ capacity to develop and benefit from AI technologies. She highlighted positive examples, noting that “Last year, the Nobel Prize for Chemistry was awarded to the developers of AlphaFold” and that this system “modeled over 200 million proteins and made it freely available to accelerate drug discovery.”
### The Governance Challenge: Beyond Principles to Practice
Martin used a maritime metaphor to explain current governance limitations, stating that while frameworks like the UN’s Pact for the Future and Global Digital Compact provide valuable guidance as a “compass,” they are insufficient alone: “a compass can’t move a ship. It can only point it in the right direction.” She argued that practical governance mechanisms are essential to translate principles into operational safeguards.
### Three Pillars of Effective AI Governance
Martin outlined three critical elements for effective AI governance, describing them as the “captain’s wheel” needed to steer AI progress.
#### Pillar 1: Inclusion
Martin highlighted that more than 100 countries have no meaningful voice in global AI governance discussions, despite 170 countries being represented at the summit. She emphasized: “we cannot outsource trust and we cannot expect countries to implement safeguards that they had no role in designing and that don’t fit their local context.”
#### Pillar 2: Capacity Building
This pillar encompasses both technical infrastructure (access to compute power, data centers, and AI systems) and human capacity (the ability to make informed decisions about AI). Martin announced the launch of the AI Skills Coalition, a collaborative initiative between the ITU and partners to provide targeted support for countries developing AI governance capabilities.
#### Pillar 3: Technical Standards
Martin argued that technical standards are essential for translating high-level commitments into operational safeguards. She noted recent consultations “at the Open Dialogue on AI testing” and “workshop on trustworthy AI testing and validation” that revealed significant challenges in developing applicable methodologies across different contexts.
## Multi-stakeholder Responsibilities
Martin outlined distinct roles for different stakeholder groups:
– **Governments**: Enacting enabling laws, protecting rights, and investing in digital public services. She cited Estonia’s approach: “back in the 90s you actually earmarked 1 percent of state funding for IT transforming Estonia into the most one of the most or the most advanced digital society in the world”
– **Industry**: Responsible AI development aligned with governance frameworks
– **Academia**: Independent evaluation and testing capabilities
– **Civil Society**: Advocacy ensuring frameworks reflect diverse societal needs
– **UN System**: Coordination and promoting universal values
## Concrete Commitments and Outcomes
Martin announced several specific initiatives:
– Continuation of consultations on AI testing and trustworthy AI validation beyond the summit
– Launch of an annual UN AI activities report, with Martin noting the UN system’s increased AI integration across its work
– The AI Skills Coalition launch to address capacity gaps
– Ongoing multi-stakeholder collaboration on AI standards and testing methodologies
## Practical Examples and Evidence
Martin provided evidence of the summit’s effectiveness, describing how “a government representative at the lunch roundtable specifically requested support from peers, leading to concrete offers of assistance.” She also referenced discussions about “identifying sources of untapped open source communities in developing countries” and specific policy areas including “deep fakes to access to compute power to red teaming always grounded in scientific observation.”
## Vision for Collaborative Governance
The presentation concluded with a collaborative vision, quoting Minister George from Ghana: “don’t look east don’t look west. But ladies and gentlemen we need to look forward together.” Martin reinforced this theme, suggesting that while countries don’t need to “sail in the same ship” or “go at the same speed,” they should “navigate the same oceans together by the same compass under the same stars.”
## Implications and Next Steps
Martin’s framework moves beyond theoretical debates to focus on practical implementation challenges. The three-pillar approach provides a roadmap for developing effective AI governance mechanisms while acknowledging diverse contexts and capabilities. The emphasis on ongoing consultations and capacity building recognizes AI governance as a continuous process requiring sustained attention and resources.
The commitment to annual reporting and the AI Skills Coalition provides concrete mechanisms for accountability and support. Most significantly, the presentation reframes AI governance from a competitive challenge to a collaborative opportunity, suggesting that AI benefits can be shared globally through inclusive, multi-stakeholder processes.
## Conclusion
Martin’s keynote address provided a practical framework for AI governance that balances universal principles with local contexts. The three-pillar approach of inclusion, capacity building, and technical standards offers a pathway from high-level commitments to operational implementation. By emphasizing collaboration over competition and providing concrete mechanisms for support and accountability, the presentation established a foundation for continued progress in global AI governance.
Session transcript
Doreen Bogdan Martin: Thank you. And we now have a chance together to reflect on AI governance with someone who has a unique and incredible view on where everything is going. Please join me in welcoming to the stage the Secretary General of the ITU. It’s Doreen Bogdan Martin. Your Excellency, Mr. Karis, President of Estonia, Excellencies, colleagues, friends. Transformation is a word that we hear often in the technology world. But when it comes to artificial intelligence, I believe we truly are in a transformational moment. The global AI race is well underway, sparking fierce competition between companies, countries, regions, reshaping diplomacy, economics, the labor markets, and exposing critical gaps in the capacity to develop, to deploy, and to benefit from AI. As this technology evolves amid fears it will overtake human ingenuity. The challenge is not whether to govern AI, but how to ensure governance steers it in the right direction. We know this is possible because we’ve already caught a glimpse of it. Last year, the Nobel Prize for Chemistry was awarded to the developers of AlphaFold, an AI system that predicts the complex 3D structure of proteins. For decades, modeling a single protein took several years. AlphaFold has now modeled virtually every protein known to science, more than 200 million. And those scientists made their work freely available, helping researchers worldwide accelerate drug discovery, transform medicine, and help the world to better understand. This was the building blocks of life that they shared with the world. That’s the kind of AI future I think we all want. One where AI opens new frontiers of scientific discovery on Earth and in space. Ladies and gentlemen, this is why we’re here. Because this transformative AI moment demands more than admiration or alarm. It demands dialogue and concerted action on inclusive, forward-looking governance that drives innovation and builds public trust. Governance that minimizes the risks and leaves no one behind. Last year, we heard a few times this week. Last year, the United Nations at the General Assembly adopted the Pact for the Future and the Global Digital Compact, complementing guidance that was offered by the World Summit on the Information Society. And we’re here also doing the plus 20 review. And these frameworks actually serve as our compass. A compass for a more equitable, rights-based AI future. But a compass can’t move a ship. It can only point it in the right direction. To steer AI progress towards shared benefits, we need governance mechanisms that are practical, that are inclusive, and that are rooted in real-world implementation. These governance mechanisms form our captain’s wheel. And here, let me thank our captains. We have our captains of today’s AI Governance Dialogue, our distinguished co-chairs. We have His Excellency, Engineer Majed al-Mezmar, the Director General of the Telecommunications and Digital Government Regulatory Authority of the UAE. It’s a long name. We also have Madame Anne Bouvereau, my friend, France’s Special Envoy for Artificial Intelligence. And welcome to you, Anne. And as we continue today’s discussions, I would like you to keep in mind three key elements. Three key elements that I think can actually help us to propel AI governance for good forward. First, inclusion. We keep hearing this word, inclusion. Too many countries, more than 100, have no meaningful voice in global AI governance discussions. While it’s encouraging to see more of these discussions taking place, from Bletchley Park to Seoul to Paris, and more recently, Kigali, the global reach of the United Nations can help make AI governance inclusive, or as inclusive as it possibly can be. This week, we’re proud to have participants from 170 countries. Did I get that right, Fred? 170 countries here at this year’s summit. And their perspectives are essential in designing governance mechanisms that truly reflect global realities. Not just high resource context, but communities. Communities that are navigating limited infrastructure, low trust, and high stakes. Many governments also lack the resources to engage in, let alone shape, their own AI futures. And I think you would agree with me, ladies and gentlemen, that must change. That brings me to my second point. Capacity. Capacity is linked to being connected to infrastructure, of course. And that includes access to compute, data centers, and other infrastructure for artificial intelligence. But capacity is also about people. It’s about people and their ability to make informed decisions. And that’s why we need to equip policymakers, public administrators, especially in developing countries, with the skills to assess, procure, and deploy AI systems. That’s why ITU and our partners launched the AI Skills Coalition. And the third and the final element that I think can help steer the AI revolution in the right direction is standards. Because principles and declarations alone are not enough. We need technical standards that translate high level commitments into operational safeguards. That’s why earlier this week we held consultations at the Open Dialogue on AI testing and we also had a workshop on trustworthy AI testing and validation. And I think it’s fair to say that these gatherings revealed an urgent need for multi-stakeholder collaboration in two key areas of action. Promoting knowledge and knowledge exchange I should say on AI standards and bringing capacity gaps and methodologies for testing AI systems and models. ITU is ready to continue convening these consultations beyond AI for good beyond this summit and we look forward to doing that with you. Because we cannot leave AI governance to chance. We cannot outsource trust and we cannot expect countries to implement safeguards that they had no role in designing and that don’t fit their local context. Bringing these three key elements together the inclusion the capacity and the standards is what coordinated steering looks like. We saw this in action at today’s roundtable. We had a lunch roundtable conversation where participants highlighted the importance of identifying sources of untapped open source communities in developing countries to broaden inclusion. And using policy tools to deliver on specific areas from deep fakes to access to compute power to red teaming always grounded in scientific observation. And actually one government one government representative I should say that was seated at the table he requested support from his peers at that table. And I thought about it and I thought this is a powerful reminder that when we bring the right people together dialogue actually goes beyond discussion to become a catalyst a catalyst for real cooperation and concrete action and hope. And I think we could all use a little hope. Ladies and gentlemen governance is a shared multi stakeholder responsibility. Everyone in the generation has a part to play and we do need all hands on deck. Governments can lead in enacting the enabling laws and protecting rights and in investing in digital public services. And here Mr. President I want to give a big shout out to the Estonian government because back in the 90s you actually earmarked 1 percent of state funding for IT transforming Estonia into the most one of the most or the most advanced digital society in the world. Industry also has a big role to play to develop and deploy artificial intelligence responsibly and to be transparent about high risk systems. Academia and the technical community can help to evaluate those models stress tests assumptions and illuminate the blind spots. Civil society can raise concerns expose harms and advocate for the communities that might be left out. And of course the United Nations system can continue coordinating can continue convening and keeping universal values of peace dignity and human rights at the core as we seek to leverage artificial intelligence responsibly. You will see maybe shortly on the screen our progress in terms of what we’re doing in the United Nations system. Our last edition we have this annual U.N. A.I. activities report that’s been compiled by the ITU that we’re proudly launching this afternoon and the report shows how the U.N. is integrating artificial intelligence across our work. At the end of last year we reported seven hundred and twenty nine to be exact projects in 2024 and that was up from 400 and 406 in 2023. We’re also seeing lots more engagement on artificial intelligence right across the U.N. system. We have 53 entities that are contributing to that process. Ladies and gentlemen the future is here at the A.I. for good global summit. But as the saying goes it’s not evenly distributed out there in the world. This is a transformative moment for A.I. technology. So let’s make it a transformative moment for governance too. Let this be remembered as the point the point where we turned the ship around. Not when we lost control but when we took the helm. Not when we raced in competition but when dialogue helped all the boats rise. We don’t need to sail in the same ship. We don’t even need to go at the same speed but we do we do. Ladies and gentlemen we do need to navigate the same oceans together by the same compass under the same stars. And if my friend from Ghana Minister George is in the room I’m going to share something he said the afternoon this afternoon. He said don’t look east don’t look west. But ladies and gentlemen we need to look forward together. And so with that ladies and gentlemen thank you for being here and look forward to the continued discussions. Thank you Mr. President. Thank you. Thank you.
Doreen Bogdan Martin
Speech speed
122 words per minute
Speech length
1585 words
Speech time
778 seconds
The global AI race is reshaping diplomacy, economics, and labor markets while exposing critical gaps in capacity to develop and benefit from AI
Explanation
Martin argues that AI is creating a transformational moment that goes beyond technology, fundamentally changing how nations interact diplomatically and economically. This transformation is creating fierce competition between companies, countries, and regions while revealing significant disparities in AI capabilities.
Evidence
The global AI race is sparking fierce competition between companies, countries, regions, reshaping diplomacy, economics, the labor markets
Major discussion point
AI Governance as a Transformational Moment
Topics
Development | Economic | Legal and regulatory
The challenge is not whether to govern AI, but how to ensure governance steers it in the right direction
Explanation
Martin emphasizes that AI governance is inevitable and necessary, but the critical question is implementing effective governance mechanisms. The focus should be on creating governance that drives innovation while building public trust and minimizing risks.
Evidence
governance steers it in the right direction, governance that drives innovation and builds public trust, governance that minimizes the risks and leaves no one behind
Major discussion point
AI Governance as a Transformational Moment
Topics
Legal and regulatory | Development
AlphaFold demonstrates the positive potential of AI by modeling over 200 million proteins and making the work freely available to accelerate scientific discovery
Explanation
Martin uses AlphaFold as a concrete example of beneficial AI implementation, showing how AI can solve complex scientific problems that previously took years. The key aspect is that the developers made their work freely available, enabling global collaboration and advancement in medicine and drug discovery.
Evidence
Last year, the Nobel Prize for Chemistry was awarded to the developers of AlphaFold, an AI system that predicts the complex 3D structure of proteins. For decades, modeling a single protein took several years. AlphaFold has now modeled virtually every protein known to science, more than 200 million. And those scientists made their work freely available, helping researchers worldwide accelerate drug discovery, transform medicine
Major discussion point
AI Governance as a Transformational Moment
Topics
Development | Sociocultural
More than 100 countries have no meaningful voice in global AI governance discussions, which must change
Explanation
Martin highlights a significant gap in global AI governance where over 100 countries are excluded from meaningful participation in shaping AI policies. This exclusion undermines the legitimacy and effectiveness of global AI governance frameworks.
Evidence
Too many countries, more than 100, have no meaningful voice in global AI governance discussions
Major discussion point
Inclusive AI Governance Framework
Topics
Development | Legal and regulatory
The UN’s global reach can help make AI governance inclusive, with 170 countries participating in the summit representing diverse perspectives
Explanation
Martin argues that the United Nations’ universal membership and convening power makes it uniquely positioned to create inclusive AI governance. The participation of 170 countries in the summit demonstrates the UN’s ability to bring together diverse global perspectives that are essential for effective governance.
Evidence
the global reach of the United Nations can help make AI governance inclusive, This week, we’re proud to have participants from 170 countries here at this year’s summit. And their perspectives are essential in designing governance mechanisms that truly reflect global realities
Major discussion point
Inclusive AI Governance Framework
Topics
Development | Legal and regulatory
Countries cannot be expected to implement safeguards they had no role in designing and that don’t fit their local context
Explanation
Martin emphasizes that effective AI governance requires local ownership and contextual relevance. Imposing externally designed safeguards without considering local realities and involving affected countries in the design process is both unfair and impractical.
Evidence
we cannot expect countries to implement safeguards that they had no role in designing and that don’t fit their local context
Major discussion point
Inclusive AI Governance Framework
Topics
Legal and regulatory | Development
Capacity includes access to compute, data centers, and AI infrastructure, but also people’s ability to make informed decisions
Explanation
Martin defines capacity building in AI governance as having both technical and human dimensions. While physical infrastructure like computing power and data centers is important, equally crucial is developing human capacity for understanding and decision-making around AI systems.
Evidence
Capacity is linked to being connected to infrastructure, of course. And that includes access to compute, data centers, and other infrastructure for artificial intelligence. But capacity is also about people. It’s about people and their ability to make informed decisions
Major discussion point
Capacity Building for AI Implementation
Topics
Development | Infrastructure | Sociocultural
Policymakers and public administrators, especially in developing countries, need skills to assess, procure, and deploy AI systems
Explanation
Martin identifies a critical skills gap among government officials who need to make decisions about AI systems. This is particularly acute in developing countries where resources and expertise may be limited, yet these officials must be able to evaluate and implement AI solutions effectively.
Evidence
we need to equip policymakers, public administrators, especially in developing countries, with the skills to assess, procure, and deploy AI systems
Major discussion point
Capacity Building for AI Implementation
Topics
Development | Sociocultural
The ITU and partners launched the AI Skills Coalition to address capacity gaps
Explanation
Martin presents the AI Skills Coalition as a concrete initiative to address the identified skills and capacity gaps in AI governance. This represents a practical response to the need for building human capacity in AI decision-making and implementation.
Evidence
That’s why ITU and our partners launched the AI Skills Coalition
Major discussion point
Capacity Building for AI Implementation
Topics
Development | Sociocultural
Principles and declarations alone are not enough; technical standards are needed to translate commitments into operational safeguards
Explanation
Martin argues that high-level policy commitments must be accompanied by detailed technical standards to be effective. Without specific technical standards, principles remain abstract and cannot be practically implemented as operational safeguards.
Evidence
Because principles and declarations alone are not enough. We need technical standards that translate high level commitments into operational safeguards
Major discussion point
Technical Standards and Implementation
Topics
Infrastructure | Legal and regulatory
There is urgent need for multi-stakeholder collaboration in promoting knowledge exchange on AI standards and addressing capacity gaps in testing AI systems
Explanation
Martin identifies two critical areas requiring immediate collaborative action: sharing knowledge about AI standards and building capacity for testing AI systems. This collaboration must involve multiple stakeholders to be effective and comprehensive.
Evidence
these gatherings revealed an urgent need for multi-stakeholder collaboration in two key areas of action. Promoting knowledge and knowledge exchange I should say on AI standards and bringing capacity gaps and methodologies for testing AI systems and models
Major discussion point
Technical Standards and Implementation
Topics
Infrastructure | Development | Legal and regulatory
ITU will continue convening consultations on AI testing and trustworthy AI validation beyond the summit
Explanation
Martin commits the ITU to ongoing leadership in facilitating discussions about AI testing and validation. This represents a long-term commitment to continue the work initiated at the summit and maintain momentum on technical standards development.
Evidence
ITU is ready to continue convening these consultations beyond AI for good beyond this summit and we look forward to doing that with you
Major discussion point
Technical Standards and Implementation
Topics
Infrastructure | Legal and regulatory
Governance is a shared responsibility requiring all hands on deck, with different roles for governments, industry, academia, civil society, and the UN system
Explanation
Martin emphasizes that effective AI governance cannot be achieved by any single actor but requires coordinated effort from all stakeholders. Each sector has distinct but complementary roles to play in creating comprehensive AI governance frameworks.
Evidence
governance is a shared multi stakeholder responsibility. Everyone in the generation has a part to play and we do need all hands on deck
Major discussion point
Multi-stakeholder Responsibility in AI Governance
Topics
Legal and regulatory | Development
Governments should lead in enacting enabling laws, protecting rights, and investing in digital public services, citing Estonia’s success in IT transformation
Explanation
Martin outlines the specific role of governments in AI governance, emphasizing their responsibility for legal frameworks, rights protection, and public investment. Estonia is presented as a successful model, having allocated 1% of state funding to IT transformation in the 1990s, becoming one of the world’s most advanced digital societies.
Evidence
Governments can lead in enacting the enabling laws and protecting rights and in investing in digital public services. And here Mr. President I want to give a big shout out to the Estonian government because back in the 90s you actually earmarked 1 percent of state funding for IT transforming Estonia into the most one of the most or the most advanced digital society in the world
Major discussion point
Multi-stakeholder Responsibility in AI Governance
Topics
Legal and regulatory | Development | Infrastructure
The UN system has significantly increased AI integration across its work, with 729 projects in 2024 up from 406 in 2023, involving 53 entities
Explanation
Martin provides concrete evidence of the UN system’s growing engagement with AI through quantitative data showing substantial growth in AI-related projects. The involvement of 53 entities demonstrates the breadth of AI integration across the UN system.
Evidence
At the end of last year we reported seven hundred and twenty nine to be exact projects in 2024 and that was up from 400 and 406 in 2023. We’re also seeing lots more engagement on artificial intelligence right across the U.N. system. We have 53 entities that are contributing to that process
Major discussion point
Multi-stakeholder Responsibility in AI Governance
Topics
Development | Legal and regulatory
Agreements
Agreement points
Similar viewpoints
Unexpected consensus
Overall assessment
Summary
This transcript contains only one speaker (Doreen Bogdan Martin) delivering what appears to be a keynote address on AI governance. No multi-speaker agreements can be identified as there are no other speakers presenting arguments.
Consensus level
Not applicable – single speaker presentation. The speaker presents a comprehensive framework for AI governance emphasizing inclusion, capacity building, and technical standards, but without other speakers, no consensus analysis is possible.
Differences
Different viewpoints
Unexpected differences
Overall assessment
Summary
No disagreements identified as the transcript primarily contains a single speaker’s presentation
Disagreement level
No meaningful disagreement analysis possible – the transcript consists of a keynote address by Doreen Bogdan Martin without substantive input from other speakers. While she references other participants, co-chairs, and mentions a roundtable discussion, their specific arguments and viewpoints are not presented in the transcript. The arguments list only contains positions from one speaker, making disagreement analysis impossible. This appears to be a consensus-building presentation rather than a debate format.
Partial agreements
Partial agreements
Similar viewpoints
Takeaways
Key takeaways
AI governance requires a three-pillar approach: inclusion (ensuring all countries have a voice), capacity building (infrastructure and skills development), and technical standards (translating principles into operational safeguards)
The challenge is not whether to govern AI, but how to steer it in the right direction, with AlphaFold serving as an example of beneficial AI that accelerates scientific discovery
AI governance is a shared multi-stakeholder responsibility requiring coordinated action from governments, industry, academia, civil society, and the UN system
More than 100 countries currently lack meaningful participation in global AI governance discussions, highlighting the need for more inclusive frameworks
Technical standards and operational safeguards are essential because principles and declarations alone are insufficient for effective AI governance
The UN system has significantly expanded its AI integration, with 729 projects in 2024 compared to 406 in 2023, involving 53 entities
Resolutions and action items
ITU will continue convening consultations on AI testing and trustworthy AI validation beyond the current summit
Launch of the annual UN AI activities report compiled by ITU to track progress across the UN system
Continued operation of the AI Skills Coalition launched by ITU and partners to address capacity gaps
Ongoing multi-stakeholder collaboration in two key areas: promoting knowledge exchange on AI standards and addressing capacity gaps in AI testing methodologies
Unresolved issues
How to practically ensure that over 100 countries without meaningful voice in AI governance can be effectively included in decision-making processes
Specific mechanisms for countries with limited resources to engage in and shape their own AI futures
How to balance the need for universal standards with local context requirements that vary significantly across countries
Concrete methods for translating high-level AI governance principles into operational safeguards that work across diverse technological and economic contexts
How to address the uneven global distribution of AI benefits and ensure equitable access to AI infrastructure and capabilities
Suggested compromises
Countries don’t need to ‘sail in the same ship’ or ‘go at the same speed’ but should ‘navigate the same oceans together by the same compass under the same stars’ – allowing for different approaches while maintaining shared principles
Focus on dialogue and cooperation rather than pure competition in the global AI race, with the goal of helping ‘all boats rise’
Balancing the need for universal AI governance frameworks with recognition that safeguards must fit local contexts and involve local participation in their design
Thought provoking comments
The challenge is not whether to govern AI, but how to ensure governance steers it in the right direction.
Speaker
Doreen Bogdan Martin
Reason
This reframes the entire AI governance debate by moving beyond the binary question of regulation versus non-regulation to focus on the quality and direction of governance. It acknowledges that AI governance is inevitable and necessary, shifting the conversation to implementation strategies rather than fundamental debates about necessity.
Impact
This comment established the foundational premise for the entire discussion, moving the conversation away from theoretical debates about AI regulation toward practical implementation challenges. It set the tone for a solutions-oriented dialogue focused on ‘how’ rather than ‘whether’ to govern AI.
But a compass can’t move a ship. It can only point it in the right direction. To steer AI progress towards shared benefits, we need governance mechanisms that are practical, that are inclusive, and that are rooted in real-world implementation.
Speaker
Doreen Bogdan Martin
Reason
This metaphor powerfully illustrates the gap between having principles/frameworks and actually implementing effective governance. It highlights that existing UN frameworks like the Global Digital Compact are insufficient without actionable mechanisms, introducing a critical distinction between guidance and execution.
Impact
This metaphor became a central organizing principle for the discussion, leading directly into the introduction of the three key elements (inclusion, capacity, standards) as the ‘captain’s wheel.’ It shifted the conversation from celebrating existing frameworks to acknowledging their limitations and the need for practical implementation tools.
Too many countries, more than 100, have no meaningful voice in global AI governance discussions.
Speaker
Doreen Bogdan Martin
Reason
This statistic starkly quantifies the exclusion problem in AI governance, revealing that the majority of the world’s countries are effectively shut out of decisions that will profoundly impact their futures. It challenges the legitimacy of current AI governance processes and highlights a fundamental democratic deficit.
Impact
This comment provided concrete evidence for why inclusion must be the first pillar of effective AI governance, justifying the UN’s role in AI governance discussions. It added urgency to the inclusion argument by showing the scale of exclusion and helped explain why existing governance mechanisms may be inadequate.
We cannot outsource trust and we cannot expect countries to implement safeguards that they had no role in designing and that don’t fit their local context.
Speaker
Doreen Bogdan Martin
Reason
This comment addresses a fundamental flaw in top-down governance approaches, highlighting that trust cannot be imposed externally and that context-specific solutions are essential. It challenges the assumption that universal AI governance solutions can be developed by a few actors and applied globally.
Impact
This insight reinforced the importance of inclusive governance processes and provided a philosophical foundation for why broad participation is not just morally right but practically necessary. It connected the inclusion argument to effectiveness, showing that exclusive governance processes are likely to fail even on their own terms.
And actually one government representative I should say that was seated at the table he requested support from his peers at that table. And I thought about it and I thought this is a powerful reminder that when we bring the right people together dialogue actually goes beyond discussion to become a catalyst for real cooperation and concrete action and hope.
Speaker
Doreen Bogdan Martin
Reason
This anecdote provides concrete evidence that inclusive dialogue can produce immediate, practical results. It demonstrates how bringing diverse stakeholders together can create opportunities for peer-to-peer support and collaboration that wouldn’t exist in more exclusive settings.
Impact
This real-world example validated the entire premise of inclusive AI governance by showing tangible results from the summit’s approach. It shifted the conversation from theoretical arguments about inclusion to concrete evidence of its effectiveness, adding credibility to the broader governance framework being proposed.
Don’t look east don’t look west. But ladies and gentlemen we need to look forward together.
Speaker
Minister George (Ghana), quoted by Doreen Bogdan Martin
Reason
This quote transcends geopolitical divisions and competition narratives that often dominate AI discussions. It suggests that the traditional East-West framing of AI competition is counterproductive and that a collaborative, future-oriented approach is necessary for addressing global AI challenges.
Impact
This comment provided a powerful concluding message that reinforced the collaborative theme throughout the speech. It offered an alternative to competitive nationalism in AI development and supported the argument for multilateral cooperation, serving as a memorable call to action for unified global governance.
Overall assessment
These key comments shaped the discussion by systematically building a case for inclusive, practical AI governance that moves beyond existing frameworks. Bogdan Martin’s strategic use of metaphors (compass vs. captain’s wheel), concrete statistics (100+ excluded countries), and real-world examples (the government representative requesting support) created a compelling narrative arc that progressed from problem identification to solution framework to evidence of effectiveness. The comments worked together to reframe AI governance from a technical or competitive issue to a fundamentally collaborative challenge requiring broad participation and practical implementation mechanisms. The discussion’s impact was enhanced by grounding abstract governance concepts in tangible examples and memorable metaphors, making complex policy arguments accessible and actionable for a diverse international audience.
Follow-up questions
How can we identify and tap into untapped open source communities in developing countries to broaden inclusion in AI governance?
Speaker
Participants at the lunch roundtable conversation
Explanation
This was highlighted as important for making AI governance more inclusive globally, particularly for countries that currently lack meaningful voice in AI governance discussions
What specific policy tools can be developed and implemented to address deep fakes, access to compute power, and red teaming?
Speaker
Participants at the lunch roundtable conversation
Explanation
These were identified as specific areas requiring policy solutions, grounded in scientific observation, to ensure effective AI governance
How can we develop methodologies for testing AI systems and models to address current capacity gaps?
Speaker
Participants in AI testing consultations and trustworthy AI testing workshop
Explanation
The consultations revealed an urgent need for multi-stakeholder collaboration in developing testing methodologies, particularly for countries with limited technical capacity
How can knowledge exchange on AI standards be promoted more effectively across different countries and contexts?
Speaker
Participants in AI testing consultations and trustworthy AI testing workshop
Explanation
This was identified as a key area requiring multi-stakeholder collaboration to ensure AI standards can be implemented globally
What specific support mechanisms can be established for countries requesting assistance with AI governance implementation?
Speaker
One government representative at the lunch roundtable
Explanation
A government representative specifically requested support from peers, highlighting the need for concrete cooperation mechanisms between countries at different stages of AI development
Disclaimer: This is not an official session record. DiploAI generates these resources from audiovisual recordings, and they are presented as-is, including potential errors. Due to logistical challenges, such as discrepancies in audio/video or transcripts, names may be misspelled. We strive for accuracy to the best of our ability.