Coalition for Decentralized Healthcare and Life Sciences (CDHLS)

A successful first Global Dialogue on AI Governance should produce an actionable roadmap that clearly identifies the most important AI risks, sets shared global governance priorities, and outlines practical next steps for countries, organizations, and institutions worldwide. The roadmap should focus on near-term actions while also creating a long-term foundation for responsible AI governance globally. A critical factor for success is strong multi-stakeholder inclusion. This means ensuring meaningful participation from developing countries, civil society, academia, industry, and other relevant groups. Broad representation from different regions and sectors would create a more balanced, credible, and legitimate global approach, while helping ensure governance solutions reflect diverse needs and realities fairly. The Dialogue should also make progress toward interoperable AI governance by promoting alignment on common principles such as AI safety, transparency, accountability, fairness, and human oversight. Shared principles can help reduce regulatory fragmentation across countries, make cross-border cooperation easier, and support responsible innovation while building trust in AI systems worldwide today. Another essential outcome is a clear commitment to ethical, trustworthy, and human rights-based AI. This should include safeguards such as auditability, oversight mechanisms, risk management, and responsible use of AI systems so that technology remains aligned with public interest, dignity, fairness, and core human values for everyone globally. The Dialogue should further promote measurable capacity-building efforts, especially for countries with limited resources. This includes sharing knowledge, improving access to AI tools and computing resources, developing skills, and investing in digital infrastructure. Such actions would help bridge the global digital divide and ensure more equitable participation in the AI economy worldwide. In addition, the Dialogue should emphasize scalable and interoperable digital infrastructure to make governance practical worldwide. Common technical standards, reference architectures, and shared platforms can help translate agreed principles into real implementation across countries, sectors, and institutions effectively and consistently over time with practical measurable outcomes for all stakeholders globally. Finally, success depends on establishing clear follow-up mechanisms such as milestones, working groups, regular reporting, and accountability structures. Ongoing coordination, informed by scientific evidence and practical experience, will ensure the Dialogue becomes a meaningful platform delivering sustained global action and benefits.

• Safe, secure and trustworthy AI
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Transparency, accountability, and human oversight
• Protection and promotion of human rights

1. Safe, secure and trustworthy AI AI systems are being increasingly used in some of the high-impact domains (but not limited) to healthcare, finance, aviation and public services. Ensuring safety and security helps prevent harmful outcomes, reduces the risk of misuse, and builds public confidence in using and adopting AI technologies. Especially in the regulatory industries, reliability, robustness, and clear validation processes, is critically important to build and ensure trustworthy AI. AI systems are increasingly deployed in high-impact domains such as healthcare, finance, aviation, and public services. Ensuring safety, robustness, and security reduces risks, prevents misuse, and builds public trust. Reliable validation and verification processes are essential, particularly in regulated sectors. 2. Social, economic, ethical, cultural, linguistic and technical implications of AI Without careful consideration, especially training AI models and algorithms, AI systems can reinforce inequalities, ignore underrepresented populations, or completely fail to account for cultural and linguistic diversity. Addressing these implications will ensure AI benefits are distributed fairly and that systems are inclusive and context-aware. AI systems can reinforce inequalities or overlook underrepresented populations if not carefully designed. Addressing these dimensions ensures inclusivity, fairness, and cultural relevance, enabling equitable distribution of AI benefits. 3. Protection and promotion of human rights AI systems shall be enforced to respect human rights such as their privacy, non-discrimination, and freedom of expression. Implementing appropriate human rights principles strictly into the governance of AI helps to prevent harm, ensures ethical use, and aligns AI development with international legal standards. AI must respect privacy, non-discrimination, and freedom of expression. Embedding human rights into governance frameworks ensures ethical use and alignment with international legal standards. 4. Transparency, accountability, and human oversight This is necessary to ensure that AI systems remain understandable and controllable. AI transparency allows involved stakeholders and regulatory agencies to understand how decisions are made which would help in developing products specifically healthcare and pharmaceuticals. Accountability identifies the organizations and individuals who are responsible for AI outcomes. Human oversight ensures timely and critical decisions are appropriately reviewed by human experts especially relevant to semi- or fully-autonomous AI systems such as Agentic AI systems. Transparency enables understanding of AI decisions, while accountability ensures responsibility for outcomes. Human oversight remains essential, particularly in high-stakes or autonomous systems, to ensure appropriate intervention and control.

One major area is the environmental and energy cost of AI and broader sustainability impacts. Another key concern is the risk from advanced AI systems, such as manipulation of human behavior, loss of autonomy, long-term safety risks, and growing dependence on automated decision-making systems worldwide today. Another major issue is the lack of globally coordinated technical infrastructure to support governance implementation. While principles are widely discussed, there is limited focus on consistent execution across systems and jurisdictions. The rise of autonomous and agentic AI systems increases challenges in verification, validation, cybersecurity, and continuous monitoring. Secure-by-design approaches are therefore essential. There are also significant gaps in accountability, measurement, and law enforcement mechanisms globally. Unequal access to computing infrastructure, skills, and data creates digital dependency, especially for developing countries. Local data sovereignty and market concentration through monopolies are also growing concerns that require coordinated global attention. There is also a growing gap between policy principles and engineering practice. Stronger collaboration among policymakers, researchers, and technical experts is needed to make governance practical, scalable, and enforceable. Decentralized infrastructure may help bridge this gap. Decentralized ledgers, verifiable computation, and protocol-based compliance can create tamper-resistant audit trails for model provenance, training data sources, and decision logs. These tools can improve trust and independent verification. Social inclusion must remain a priority. Governance discussions shall address barriers affecting women, youth, children, caregivers, linguistic minorities, artists, persons with disabilities, and developing nations. Additional issues include protection of children in AI interactions, governance of digital identities after death, and transparency over how AI systems shape knowledge, definitions, and public understanding. Finally, governance should recognize new collaborative decentralized models that reward global contributors for solving underserved challenges, such as rare diseases, neglected conditions, and low-resource healthcare needs where traditional commercial incentives are often insufficient today.

Governance gaps are creating both serious challenges and important opportunities. A major challenge is fragmentation across regulatory and technical approaches, that increases compliance complexity, and slows innovation. Many organizations, especially in developing regions, also lack the technical capacity, computing infrastructure, and skilled workforce needed to implement governance requirements effectively. In sectors such as finance, education, and the creative economy, AI is advancing faster than governance frameworks. This creates uncertainty around accountability, explainability, IP, privacy, safety, and responsible deployment. In many countries, dependence on imported AI systems risks widening economic and technological divides while weakening local innovation ecosystems over time significantly. Healthcare provides a clear example of both challenge and opportunity. Organizations need to protect IP and patient privacy while also having cross-institutional data sharing for robust AI training. Privacy-preserving techniques such as federated learning, where models train across institutions without raw data leaving organizational boundaries, offer a promising technical path forward. However, governance frameworks need to evolve to support these emerging techniques. Decentralized and open approaches to AI development create new opportunities. Open-source models combined with decentralized infrastructure can lower costs, expand access, and help deliver specialized AI tools to smaller hospitals, rural providers, and underserved communities. This can improve equity and broaden the benefits of innovation. Other key challenges include exclusion of underrepresented groups, weak support for local talent, and lack of clear global standards. However, these gaps also create opportunities to build shared tools, common frameworks, cloud-based compliance platforms, and scalable monitoring systems that can be used across jurisdictions. Strengthening international cooperation and investing in technical capacity-building can help bridge these divides. With stronger local ecosystems, inclusive participation, and investment we can create more consistent, practical, and equitable AI governance that supports innovation while protecting public interests worldwide.

The AI Dialogue can help in advancing international cooperation on AI governance by helping close the gap between rapid AI innovation and slower governance responses. Here governments, industry, academia, civil societies, and technical experts can work together to build shared guardrails and common minimum standards for responsible AI development. The Dialogue can align national and regional AI governance frameworks while respecting legal and cultural differences. By promoting shared principles like safety, transparency, accountability, fairness, and human oversight, it can reduce fragmentation, improve interoperability between jurisdictions, and enable smoother cross-border innovation, trade, and collaboration in global AI markets. It can build trust and promote knowledge exchange by sharing best practices, technical expertise, and policy experience across countries and sectors. This is particularly valuable for nations at different stages of AI readiness, especially developing countries to participate effectively in the global ecosystem. The Dialogue can bridge policy and implementation by supporting practical frameworks, technical guidelines, common standards, and shared digital infrastructure. Through reference models, implementation roadmaps, and scalable compliance tools, it can help countries and organizations turn broad governance principles into measurable, deployable, and effective real-world solutions at scale globally. The Dialogue can also create momentum for stronger international cooperation through reporting obligations, regular progress reviews, working groups, and continuous follow-up mechanisms to track commitments, compare progress over time, and maintain accountability. In addition, it can promote inclusive global governance by ensuring meaningful participation from developing countries and underrepresented communities, while also addressing harms affecting children, creators, vulnerable groups, and smaller economies. It can support open, trustworthy AI systems and encourage domestic AI ecosystem development where needed. Overall, the AI Dialogue can become a practical, trusted, and action-oriented platform that coordinates global efforts, supports equitable participation, and advances effective international cooperation on AI governance for the long-term benefit of all societies worldwide.

The AI Dialogue should build upon existing international initiatives, governance frameworks, partnerships, and technical ecosystems that are already addressing responsible AI development. Important examples include the EU AI Act, UN resolutions, national sector-based governance approaches, and multilateral efforts led by organizations such as (but not limited to) the UN, UNESCO, World Bank .These bodies provide valuable experience in policy coordination, standards development, sustainability, education, and international cooperation worldwide. The Dialogue should also connect with issue-specific initiatives such as the Internet Governance Forum, Responsible AI research programs, and existing public consultation mechanisms. These initiatives already bring together governments, civil society, academia, and technical communities, and can offer practical lessons on inclusive participation, accountability, and implementation. Regional and national examples are equally valuable. These include the FAIR guidelines, the UN Sustainable Development Goals framework, and data spaces initiatives for secure data sharing. These examples show how governance can be linked to digital infrastructure, economic development, inclusion, and public service delivery. The added value of the AI Dialogue would be its ability to bring these fragmented efforts together into one coordinated global platform. It can reduce duplication, identify common priorities, and accelerate progress through shared learning. It can also create stronger links between policy and implementation by encouraging global engineering collaborations where academia, industry, governments, and civil society co-develop reusable tools, datasets, system components, and governance solutions. In addition, the Dialogue should connect with open-source ecosystems and technical communities that support trustworthy AI systems, shared tools, interoperable standards, and transparent development practices. By connecting existing initiatives rather than replacing them, the Dialogue can strengthen trust, improve interoperability, scale best practices, and ensure that AI governance remains practical, inclusive, and globally relevant. This would help transform separate efforts into a more coherent and effective international governance ecosystem for the long-term benefit of all.

Different stakeholders can contribute to the AI Dialogue by sharing expertise, practical experiences, policy lessons, technical solutions, and community perspectives. Governments can contribute regulatory approaches, public policy experience, and implementation lessons from national and regional governance systems. Industry, startups, and innovators can help test practical governance frameworks, scalable compliance tools, and real-world deployment models. Academia, researchers, and technical experts can provide evidence-based proposals, standards, lifecycle controls, and independent scientific input for responsible AI development worldwide today. Civil society, think tanks, consortium organizations, and community representatives can ensure that social, cultural, ethical, and human-centered concerns are fully represented. Their participation is important to reflect the needs of vulnerable groups, children, workers, and communities directly affected by AI systems. To make the Dialogue effective and inclusive, the format shall support written submissions, open consultations, hybrid participation, multilingual accessibility, and remote engagement so that people worldwide can participate regardless of location. Representation should include developing countries, underrepresented communities, smaller economies. Capacity-building initiatives should also support broader participation from developing regions to ensure equitable representation and meaningful contribution. The structure should encourage multidisciplinary working groups (not limited to) policymakers, engineers, researchers, and domain experts. This ensures that governance proposals are technically feasible, practical, and relevant to real-world needs. Balanced leadership roles, co-chair opportunities, flexible schedules, and anonymous feedback channels can further improve participation and trust. In addition, the Dialogue should include measurable indicators, progress reviews, shared data initiatives, and ongoing collaboration mechanisms that continue beyond a single event. Governments, industry, academia, and civil society should work together through continuous engagement rather than one-time discussions. By combining inclusive participation with technical depth and sustained follow-up, the AI Dialogue can become a practical global platform that connects policy discussions with implementation, strengthens cooperation, and supports fair, effective, and human-centered AI governance worldwide.

These include stakeholders from developing countries, smaller economies,regional innovators, small enterprises, public-interest, humanitarian organizations, nonprofits, and countries with limited technical capacity. Many of these groups are directly affected by AI systems but often have limited access to international governance forums, and decision-making processes today worldwide. Socially underrepresented groups also include women, youth, children, persons with disabilities, linguistic and cultural minorities, and communities most directly impacted by AI systems. Their lived experiences are essential for designing governance that is fair, inclusive, and responsive to real societal needs. Without their participation, policies may fail to address important risks, inequalities, and unintended harms. Ordinary users, families affected by new AI-related harms, and practitioners focused on real-world implementation should also have a stronger voice in governance discussions. In addition, technical practitioners such as software engineers, system architects, cybersecurity specialists, and implementation experts are sometimes underrepresented, even though they play a critical role in building, deploying, and maintaining AI systems. Their involvement is essential to ensure governance frameworks are practical, scalable, secure, and technically feasible. These communities can be better included through proactive outreach, stronger regional representation, multilingual participation, and hybrid meeting formats that allow remote engagement. By providing travel grants, connectivity support, and accessible consultation processes can reduce participation barriers. Capacity-building programs should help developing regions and smaller organizations engage meaningfully in governance processes. Leadership opportunities, balanced representation in working groups, and continuous engagement mechanisms beyond one-time events are also important. Governments, International Organisations, and private institutions should actively seek input from underrepresented groups rather than waiting for participation. By including these voices in a sustained and meaningful way, global AI governance can become more legitimate, contextually relevant, practical, and equitable, while better reflecting the diverse needs of societies worldwide and future generations globally.

Innovative collaborative formats such as design labs, simulation exercises, and "policy-to-prototype" workshops can be especially effective. These approaches allow governments, industry, academia, civil society, and technical experts to test governance ideas in realistic scenarios, explore trade-offs, and co-develop practical solutions that can be implemented in the real world. Operational deployment testing and action-oriented forums can further connect policy discussions with actual systems and use cases. Some of the engagement ways to include structured follow-up roadmaps with milestones, annual benchmark reporting, and quarterly thematic progress reviews can create accountability and help track whether commitments are being translated into action. These mechanisms can also support learning over time by identifying what is working, where gaps remain, and which areas need stronger cooperation or investment. Hybrid participation models combining in-person and virtual engagement are essential to broaden accessibility and inclusion. Multilingual participation, interpretation services, flexible scheduling, childcare support, remote access, and accessibility-first design can help remove barriers for stakeholders from different regions and backgrounds. Anonymous feedback systems may also encourage honest input, especially where participants face political or institutional constraints. Together, these formats can create a more participatory, action-oriented, accountable, and globally inclusive AI Dialogue that delivers practical and sustained results over time for all stakeholders worldwide.

One important approach is "secure-by-design" and "ethics-by-design," where safety, privacy, fairness, and accountability requirements are built into AI systems from the earliest stages of development which helps prevent risks before they scale widely. Open standards, shared testing frameworks, lifecycle auditability, approval controls, and continuous monitoring tools can improve transparency, reliability, and long-term accountability. Baseline indicators, benchmark reporting, and measurable oversight systems also help track progress. Cloud-based platforms can provide scalable infrastructure for development, testing, and validation, especially for resource-constrained organizations and developing countries seeking responsible AI adoption at lower cost and faster implementation timelines globally. Decentralized infrastructure such as Federated learning allows cross-institutional AI training without sharing raw data, helping protect privacy and data sovereignty. Verifiable audit trails can record training data sources, model versions, and evaluation results. Decentralized AI agent orchestration can support secure AI operations across organizational boundaries and legacy systems. Open licensing for publicly funded or incentivized AI can broaden access to foundational capabilities. Existing policy examples include the EU AI Act, shared risk classification systems, sector-by-sector regulation, anti-monopoly agreements, regulatory sandboxes, and contextual deployment frameworks tailored to different sectors. Binding international conventions with clear red lines can also strengthen accountability where voluntary measures are insufficient. Inclusive governance models are equally important. Examples include FAIR principles Fairness, Accountability, Inclusiveness, and Responsibility along with access, trust, and inclusion frameworks. Governance shall also be tested in real-world settings such as refugee camps, floods, clinics, and schools to ensure practical effectiveness. Other promising practices include shared interoperable data spaces, AI tools integrated into policymaking, creator credit and compensation frameworks, gender-disaggregated indicators, and innovation ecosystems designed with women included from the beginning. Together, these policies, platforms, and practices show that effective AI governance must be technical, measurable, inclusive, scalable, and grounded in real-world implementation across societies and sectors worldwide.