International Diabetes Federation (IDF)

A successful first Global Dialogue on AI Governance would achieve three key outcomes. First, it should establish a shared baseline of globally relevant principles. These should affirm that AI governance must be human-centred, rights-based, safe, transparent, accountable, and equitable. At the same time, the Dialogue should recognise that governance cannot be one-size-fits-all, as countries differ in regulatory capacity, infrastructure, health systems, and digital readiness. Second, it should move beyond broad aspirations and define priority areas for action. Success would mean identifying practical domains for international cooperation, such as data governance, privacy, interoperability, bias and fairness, safety evaluation, transparency, and cross-border standards. In healthcare and diabetes care, for example, this means ensuring AI tools are clinically validated, context-appropriate, and accessible across diverse populations, including people in low- and middle-income settings. Third, the Dialogue should create a credible process for sustained collaboration. Governments, multilateral agencies, academia, civil society, patient communities, and responsible industry all need a seat at the table. A successful first Dialogue should not end as a one-time exchange, but should lead to structured follow-up through working groups, consultation mechanisms, and measurable next steps. From our perspective, success would also mean recognising that AI governance is not only about reducing harm, but also about enabling public good. Well-governed AI has the potential to strengthen health systems, improve access to care, support health workers, and reduce inequities. The challenge is to build trust while still enabling responsible innovation. In summary, success would mean: shared principles, actionable priorities, and an inclusive, continuing global mechanism for responsible AI governance.

• Safe, secure and trustworthy AI
• AI capacity-building
• Transparency, accountability, and human oversight
• Open-source software, open data and open AI models

From the IDF perspective, these four priorities are foundational to ensuring that AI serves people living with diabetes in a safe, equitable, and meaningful way. Safe, secure and trustworthy AI is essential because AI applications in diabetes care may influence screening, diagnosis support, glucose monitoring, insulin titration, education, and long-term self-management. Such tools must therefore be clinically reliable, secure, and appropriate for real-world use across diverse settings. AI capacity-building is a major priority for IDF because the benefits of AI will not be equitably realised unless healthcare professionals, diabetes educators, health systems, and people living with diabetes are equipped to understand and use these tools responsibly. This is especially important in low- and middle-income countries, where gaps in digital readiness and AI literacy may widen existing inequities. Transparency, accountability, and human oversight are central to IDF's vision of responsible AI in diabetes care. AI should support clinical judgement and patient empowerment, not replace the human relationship at the heart of care. Clear accountability, appropriate explainability, and meaningful human oversight are necessary to build trust and protect patients. Finally, open-source software, open data, and open AI models are important because they can help foster innovation, collaboration, and more equitable access, particularly in underserved settings. For IDF, openness can support the development of locally adaptable and context-relevant diabetes solutions, provided it is accompanied by strong safeguards for privacy, quality, safety, and responsible use. Together, these priorities reflect IDF's commitment to promoting AI that is not only innovative, but also safe, inclusive, transparent, and globally relevant in advancing diabetes care.

Yes. While the listed themes are important, several cross-cutting issues deserve stronger and more explicit attention. First, equity of access and representation should be treated as a central issue, not only as a secondary outcome of governance. In health and diabetes care, AI systems may underperform if they are trained on data that do not adequately represent diverse populations, regions, languages, socioeconomic realities, and care settings, particularly in low- and middle-income countries. Second, context-specific implementation is critical. Governance discussions often focus on high-level principles, but the real challenge is how AI is introduced into health systems, clinical workflows, and community settings. Questions of usability, affordability, interoperability, infrastructure readiness, and workforce adaptation are essential for meaningful adoption. Third, evaluation and post-deployment monitoring deserve more visibility. Governance should not end at design or regulatory approval. AI systems require continuous monitoring for safety, effectiveness, bias, unintended consequences, and performance drift in real-world settings. Fourth, AI literacy and public trust are emerging cross-cutting concerns. Healthcare professionals, patients, policymakers, and the public need the capacity to understand both the opportunities and limitations of AI. Without literacy and trust, even well-governed technologies may fail to achieve public benefit. Finally, clear accountability in multi-actor ecosystems remains an unresolved challenge. In many AI applications, responsibility is distributed across developers, deployers, health systems, regulators, and end users. Governance frameworks should better address who is accountable when harm occurs, when systems fail, or when outputs are used beyond intended settings. From an IDF perspective, these issues are highly relevant because responsible AI in diabetes care must be not only technically sound, but also equitable, implementable, continuously evaluated, and trusted across diverse global contexts.

In the diabetes sector, governance gaps in AI are already shaping both the challenges and opportunities we see globally, particularly across low- and middle-income countries. The most significant challenge is the uneven readiness of health systems to adopt AI safely and effectively. While digital tools for glucose monitoring, insulin delivery, risk prediction, screening, and patient support are expanding rapidly, governance frameworks often lag behind. There are persistent gaps in data quality, interoperability, privacy safeguards, clinical validation, and post-deployment monitoring. In many settings, AI tools are introduced without sufficient local validation or without clarity on accountability, human oversight, and appropriate use in routine care. A second major challenge is inequity. Many AI solutions are developed using data, infrastructure, and workflows from high-resource settings, which can limit their relevance and performance in other contexts. In the diabetes field, this risks widening disparities in access, affordability, and quality of care, particularly for underserved populations and regions with limited digital infrastructure. At the same time, the opportunities are considerable. AI has the potential to strengthen diabetes prevention, screening, education, self-management, and decision support at scale. It can support healthcare professionals, improve efficiency, and help extend quality care to populations that currently face workforce and access constraints. For the IDF community, this creates an important opportunity to help shape responsible governance that is globally relevant, practical, and equity-oriented. There is strong potential for international collaboration on standards, capacity-building, validation frameworks, and shared principles for trustworthy AI in diabetes care. In summary, the sector is at a pivotal moment: governance gaps create real risks, but with the right frameworks, AI can become a powerful tool for improving diabetes care and reducing inequities worldwide.

The AI Dialogue can play a valuable role as a neutral global convening platform that helps bridge fragmented approaches to AI governance. Its most important contribution would be to create a space where governments, multilateral organisations, academia, civil society, patient communities, and responsible industry can align around shared principles while respecting differences in national context and regulatory maturity. From an IDF perspective, the Dialogue can advance international cooperation in four ways. First, it can help build common language and shared principles for trustworthy AI, including safety, transparency, accountability, human oversight, equity, and inclusion. A common foundation is essential if AI governance is to be coherent across borders. Second, it can promote practical collaboration on priority issues such as data governance, interoperability, validation, bias mitigation, post-deployment monitoring, and cross-border learning. These are especially important in health, where AI tools may affect patient safety, quality of care, and access. Third, the Dialogue can help ensure that low- and middle-income countries are meaningfully included in shaping governance. International cooperation will only be credible if it reflects diverse health systems, languages, infrastructure realities, and population needs, rather than being driven only by high-resource settings. Fourth, it can support continuity and follow-through by linking discussion to working groups, technical exchanges, and measurable next steps. The value of the Dialogue will depend not only on what is discussed, but on whether it creates a durable mechanism for sustained cooperation. In summary, the AI Dialogue can help move the global community from fragmented debate toward more inclusive, practical, and internationally coordinated governance that enables innovation while protecting people and public trust.

The AI Dialogue should build on existing global frameworks rather than duplicate them. Important reference points include UNESCO's Recommendation on the Ethics of AI, which provides a universal values-based foundation for Member States; the OECD AI Principles, recently updated to reflect rapid advances including generative AI; and the WHO guidance on ethics and governance of AI for health, including its more recent guidance on large multimodal models in healthcare. It should also connect with the GPAI/OECD partnership, which supports international cooperation on trustworthy, human-centric AI. From an IDF perspective, the Dialogue should also be open to learning from sector-specific initiatives, especially in health, where governance must address safety, clinical validation, human oversight, equity, and real-world implementation. Many global and regional efforts already exist, but they are often fragmented across ethics, regulation, technical standards, and sectoral practice. The added value of the AI Dialogue would be threefold. First, it can serve as a bridging platform across existing initiatives, helping connect broad governance principles with practical sectoral realities. Second, it can promote greater coherence and interoperability across frameworks, reducing fragmentation and helping countries align without requiring identical regulatory models. Third, it can provide an inclusive multistakeholder forum that more meaningfully incorporates low- and middle-income countries, patient communities, and underrepresented sectors, including chronic disease care. In summary, the Dialogue's value lies not in starting from scratch, but in connecting existing global work, identifying gaps, and helping translate principles into more coordinated, practical, and equitable governance pathways.

Different stakeholders should contribute in ways that reflect their distinct responsibilities and expertise. Governments can share regulatory experiences and public policy priorities. UN agencies and multilateral bodies can help align global norms and ensure inclusivity. Academia and technical experts can contribute evidence, evaluation methods, and independent analysis. Civil society and patient organisations should bring lived experience, equity concerns, and public-interest perspectives. Responsible industry can provide implementation insights, practical challenges, and lessons from deployment. This multistakeholder approach is consistent with the Dialogue's aim to provide an inclusive UN platform for states and stakeholders to discuss critical AI governance issues. In terms of format, the AI Dialogue would benefit from a hybrid, structured model: 1. High-level plenary sessions to frame shared principles and strategic priorities. 2. Thematic roundtables or breakout tracks focused on concrete topics such as safety, capacity-building, health, education, equity, and open innovation. 3. Sector-focused consultations so that governance discussions can address real-world implementation challenges. 4. Written inputs and follow-up working groups to ensure continuity beyond the event itself. The current process already includes written stakeholder submissions and the inaugural Dialogue is scheduled for 6–7 July 2026 in Geneva, alongside the AI for Good Summit. From the IDF perspective, we would welcome the opportunity to contribute as a global health and diabetes stakeholder. IDF would be pleased to participate in the July 2026 Geneva meeting, and to share the perspective of the diabetes community on responsible, equitable, and trustworthy AI in chronic disease care. We believe health—and especially long-term conditions such as diabetes—should have a visible place in the Dialogue, because AI governance in this area must address safety, validation, human oversight, equity, and implementation across diverse global settings.

Yes. One of the most underrepresented perspectives in global AI governance is the medical and broader health community, especially those working in long-term, real-world care for chronic conditions such as diabetes. Global discussions are often led by governments, technology actors, and general policy experts, while the voices of clinicians, diabetes educators, patient communities, and disease-focused professional networks remain less visible. Yet these are the groups directly engaging with the benefits, risks, and implementation realities of AI in care delivery. From the IDF perspective, this gap is significant. The International Diabetes Federation represents 251 national diabetes associations in 158 countries and territories, and in 2025 established its Technology and AI Working Group to help guide responsible, equitable, and practical use of AI in diabetes care. This working group was formed precisely because the diabetes and medical communities need a stronger voice in shaping how AI is governed, evaluated, and deployed. Other underrepresented perspectives include people living with chronic diseases, healthcare workers from low- and middle-income countries, allied health professionals, and communities facing digital access barriers. Their lived realities are essential for ensuring that AI governance is not only technically sound, but also equitable, affordable, and implementable. These groups could be included through dedicated health-sector roundtables, formal participation of patient and professional organisations, regional consultations, and ongoing thematic working groups within the Global Dialogue. As the UN process moves toward the inaugural 6–7 July 2026 Geneva meeting, IDF would welcome the opportunity to participate and contribute the perspective of the global diabetes and medical community to this important dialogue.

To foster meaningful and dynamic engagement, the AI Dialogue should combine broad discussion with structured consensus-building. A particularly useful format would be a modified Delphi consensus process. This allows diverse stakeholders to contribute views in iterative rounds, respond to anonymised summaries, and gradually identify areas of convergence and divergence. The Delphi method is already widely used in health and policy settings to build robust multi-stakeholder consensus, and it would be especially valuable for complex AI governance questions where perspectives differ across sectors and regions. A strong format could include four elements. First, pre-Dialogue written submissions and digital surveys to gather priorities in advance. Second, thematic breakout sessions during the Dialogue to discuss issues such as safety, transparency, capacity-building, health, and equity. Third, Delphi-style voting rounds after these discussions to refine key principles, priority actions, and areas needing further work. Fourth, post-Dialogue working groups to carry forward the consensus into practical recommendations. This would fit well with the UN Dialogue's broader roadmap, which already includes stakeholder submissions and a formal programme for the inaugural 6–7 July 2026 Geneva meeting. Other useful engagement formats could include short sector-focused roundtables, case-based discussions, and regional consultations so that governance is informed by real implementation experiences, not only high-level principles. From the IDF perspective, this type of structured engagement would be highly valuable. As a global federation representing the diabetes community, IDF would welcome the opportunity to contribute to the July 2026 Geneva meeting and to participate in health-focused and consensus-oriented discussions on responsible, equitable, and trustworthy AI. A Delphi-informed approach would help ensure that voices from medicine, chronic disease care, and patient-centred practice are meaningfully included in the global dialogue.

From the IDF perspective, one practical example of an approach that supports effective AI governance is the multi-regional roundtable process led by the IDF AI Technology Working Committee. So far, the Committee has convened a series of regional expert roundtables across different geographies, including South Asia, Europe, MENA, and APAC, bringing together clinicians, diabetes specialists, digital health experts, researchers, and other stakeholders to discuss the responsible use of AI and technology in diabetes care. These roundtables were designed not as promotional events, but as structured listening and consensus-building exercises. Their value lies in several areas. First, they create a multistakeholder platform where real-world challenges can be discussed openly, including data quality, bias, safety, clinical validation, access, affordability, interoperability, transparency, and human oversight. Second, they ensure that governance discussions are informed by regional diversity, since the opportunities and barriers for AI in diabetes care differ significantly across health systems and resource settings. Third, they help surface shared priorities that can be taken forward into a more formal Delphi-style consensus process. One concrete outcome of this work has been the identification of 20 key need-gaps for responsible AI in diabetes care, covering issues such as data standardisation, bias auditing, external validation, reporting standards, privacy, equity, infrastructure gaps, education, and ethical accountability. These roundtables therefore function as a practical governance mechanism: they connect principle to practice, bring underrepresented clinical voices into the discussion, and help build globally relevant but context-sensitive recommendations. From our perspective, this is a useful model for AI governance: regional consultation, multistakeholder dialogue, and iterative consensus-building to guide responsible adoption.