International Foundation for Electoral Systems

A successful first Global Dialogue on AI Governance should move beyond general discussions and deliver outcomes that build trust, equity, and shared responsibility. This includes recognizing diverse cultural, linguistic, and socio-economic contexts so that global AI frameworks are not shaped solely by a handful of powerful actors, but instead reflect truly global priorities. A primary measure of success will be the creation of a formal roadmap for bridging the AI divide. The Dialogue should result in actionable commitments toward capacity-building, including access to infrastructure, data, and skills development. Promoting open-source technologies and strengthening digital public infrastructure can help countries like Sri Lanka participate more actively in the AI ecosystem rather than remaining passive consumers. These efforts must be paired with sustainable investments and partnerships that empower local innovation. From my experience working with communities and institutions in developing contexts, one of the most urgent gaps is the lack of robust legal and regulatory systems to monitor AI effectively without violating individual rights. Many developing countries face challenges in balancing innovation with safeguards due to limited technical expertise, weak enforcement mechanisms, and outdated legal frameworks. Therefore, a successful Dialogue must launch a technical assistance facility that helps these nations draft interoperable, rights-respecting AI laws. This facility should offer model rules, audit standards, and independent oversight templates, not as impositions, but as adaptable tools that prevent surveillance abuses while enabling effective governance. Furthermore, success requires the institutionalization of human rights and robust human oversight as non-negotiable foundations of AI deployment. This involves creating global mechanisms for transparency and accountability that comply with international law. The Dialogue will be a success if it aligns Member States on a shared "language" of safety and interoperability. By ensuring that AI governance reflects cultural and linguistic diversity and prioritizes ethical safeguards, the international community can prevent a fragmented regulatory landscape and ensure that AI serves the collective interests of all humanity. Just as importantly, the Dialogue should not end as a single event but evolve into an ongoing, inclusive process that continuously engages governments, civil society, and underrepresented voices.

• AI capacity-building
• Protection and promotion of human rights
• Safe, secure and trustworthy AI

The selection of these priorities is driven by the reality that AI is being deployed in an increasingly volatile global landscape. With a significant collapse in development sector funding and growing conflicts, the resource gap between nations is widening, making international coordination a survival necessity rather than a choice. 1. AI Capacity-Building In a climate of shrinking development budgets, traditional aid is becoming scarce. Capacity-building is the only sustainable way to ensure developing nations are not left in a state of "digital colonization." By focusing on skill transfer and infrastructure support, we can empower these regions to build their own solutions to local problems-such as climate-resilient agriculture or crisis management-rather than relying on expensive, imported technology they cannot maintain or control. 2. Safe, Secure, and Trustworthy AI Growing global conflicts have heightened the risk of AI being weaponized or used for disinformation. Without a unified global standard for "trustworthy" AI, the technology can exacerbate instability. Prioritizing safety ensures that AI systems are resilient against manipulation and that interoperable standards prevent a "race to the bottom" where safety is sacrificed for strategic or commercial speed. 3. Protection and Promotion of Human Rights Fragile legal systems in many developing countries are currently ill-equipped to monitor AI without violating individual liberties. In conflict-affected or resource-strained areas, there is a high risk that AI could be used for intrusive surveillance or discriminatory resource allocation. By centering human rights, the Global Dialogue ensures that international legal foundations protect the most vulnerable, even when domestic frameworks are still maturing. It transforms AI from a potential tool of control into a protected medium for civic participation and inclusion.

Yes, in my opinion, two critical cross-cutting issues are not adequately captured by the listed themes. First, the environmental sustainability of AI. While "social, economic, ethical, cultural, linguistic and technical implications" may implicitly include environmental concerns, the staggering carbon footprint of training large models, water consumption for data center cooling, and e-waste from rapid hardware obsolescence demand explicit recognition. Developing countries hosting data centers or mining critical minerals bear disproportionate environmental costs without necessarily reaping AI's benefits. This intersects with capacity-building (helping nations measure and mitigate AI's ecological impact) and open models (which can be more efficiently fine-tuned locally). Second, meaningful participation mechanisms for marginalized communities. The themes address human rights and transparency, but not the structural barriers that prevent women, indigenous peoples, rural populations, and persons with disabilities from shaping AI governance. Based on the earlier concern about weak legal systems in developing countries, even robust laws fail if affected communities cannot trigger accountability mechanisms. A missing theme is "inclusive design and grievance redress"-ensuring that feedback loops exist where rights violations occur, not just where regulators are based. This cross-cuts protection of human rights and capacity-building, but deserves its own spotlight to avoid being an afterthought. Both issues require urgent action because they represent governance gaps that no single country can close alone, and they disproportionately impact the very nations the AI Dialogue seeks to include meaningfully.

Governance gaps in AI and digital technologies are already visible across South Asia, including in Sri Lanka and Nepal, where rapid technological change is outpacing regulatory and institutional readiness. In Sri Lanka, ongoing debates around internet regulation highlight the tension between safety and freedom of expression. Proposed online safety laws intended to address misinformation and cybercrime have raised concerns about overreach and potential restrictions on civic space. At the same time, gaps in data protection and system security persist, for example, studies show that a large proportion of digital platforms share sensitive user data with limited transparency. On the other hand, the government has a solid 68-page AI strategy but only Rs. 3 billion allocated—far below the Rs. 20-40 billion needed for a single data centre . Experts warn against copying global models, emphasizing that Sri Lanka must develop its own AI solutions for local languages, plantation health, and education These challenges reflect weak enforcement capacity and evolving legal frameworks, creating risks for both privacy and trust in AI-enabled systems. In Nepal, recent protests triggered by a government-imposed social media ban illustrate how poorly designed digital governance can escalate into conflict. The ban, "introduced under regulatory justifications", was widely perceived as an attempt to suppress dissent, leading to mass protests, casualties, and eventual policy reversal. After the September 2025 protests that killed 77 people and toppled the government over social media bans, the country now faces an "AI-generated disinformation" flood. Political parties shared AI-created images of crowds inflated from 5,000 to 500,000 supporters, and deepfake videos showed candidates endorsing rivals. With over 600 complaints filed and low digital literacy where "people believe what they see," Nepal lacks both the legal framework and technical expertise to monitor this onslaught while protecting free expression. At the same time, incidents such as government website defacements by groups like 'Hacktivist Nepal' show how weak cybersecurity governance can intersect with political instability. The spread of misinformation during unrest, including false viral content, further demonstrates the risks of unregulated AI-driven information ecosystems. The opportunity lies in the fact that Nepal's judiciary has already contributed to developing UNESCO's Guidelines for AI Use in Courts, showing willingness to build governance capacity. Regarding non-state actor threats and peacebuilding, recent hacks and conflicts show asymmetrical risks are real. Terrorist organizations are already exploring AI for drone precision, automated propaganda, and deepfake identity theft. AI-enhanced cyberattacks could target critical infrastructure with minimal cost, and biological weapon development via AI tools is a growing concern. The opportunity is that global cooperation, particularly between major powers like the US and China, could establish shared safety guidelines and threat information-sharing mechanisms. Most critically, the ongoing peacebuilding efforts in both countries are threatened by governance gaps. In Nepal, misinformation "undermines the foundation of democracy" itself. In Sri Lanka, fragmented digital platforms and a low culture of data-driven decision-making hamper effective governance. The opportunity is that UNESCO and regional bodies are actively working to equip judicial actors with AI governance knowledge, ensuring justice and human rights remain priorities even as technology advances.

The AI Dialogue can play a critical role in advancing international cooperation by providing a neutral and inclusive platform where countries jointly shape AI governance norms. At present, AI development and regulation are largely driven by a few technologically advanced nations, which risks creating fragmented or uneven standards. The Dialogue helps correct this by ensuring that developing countries can contribute their perspectives, particularly on issues such as equity, access, and context-specific risks. One key role is facilitating practical cooperation and capacity-building. For example, countries like Sri Lanka are still strengthening data protection frameworks and institutional readiness to govern emerging technologies. Through the Dialogue, such countries could access technical expertise, share best practices, and build partnerships for skills development, digital infrastructure, and responsible AI adoption in sectors like public services and education. This type of cooperation ensures that AI benefits are more evenly distributed. The Dialogue is also well-positioned to promote interoperable, rights-based governance approaches. Recent developments in South Asia illustrate why this matters. In Sri Lanka, concerns around online regulation highlight the challenge of addressing misinformation while safeguarding freedom of expression. In Nepal, protests linked to digital restrictions demonstrate how governance decisions can escalate tensions if not carefully designed. These examples show the need for shared global principles—such as transparency, accountability, and human oversight—that countries can adapt locally while maintaining alignment with international human rights standards. Additionally, the Dialogue can strengthen cooperation on cross-border challenges, including cybersecurity threats, misinformation, and the misuse of AI in conflict settings. Recent incidents of cyberattacks and the spread of manipulated content in the region underscore the urgency of coordinated responses.

The AI Dialogue should build on a range of existing global, regional, and local initiatives that are already shaping AI governance and capacity-building. At the global level, frameworks such as UNESCO's Recommendation on AI Ethics provide a strong normative foundation. In particular, the Dialogue can utilize UNESCO's Readiness Assessment Methodology (RAM) and Ethical Impact Assessments to move from high-level principles to localized implementation, especially in regions like South Asia where governance systems are still evolving. Similarly, the OECD AI Policy Observatory—with its repository of 800+ national policies—offers a valuable resource to promote interoperability, ensuring that definitions of "trustworthy AI" remain compatible across contexts, from the EU to countries like Sri Lanka or Nepal. Regional frameworks also offer important lessons. Initiatives such as the ASEAN AI Guide demonstrate how voluntary guidelines can gradually transition into more formal governance approaches. The AI Dialogue can build on this "voluntary-to-formal" model as a scalable pathway for the Global South, where rigid regulatory systems may not yet be feasible. In addition, as a direct outcome of the Global Digital Compact (GDC), the Dialogue has a unique opportunity to operationalize the concept of Digital Commons, transforming open-source data and AI models into protected global public goods that are accessible and equitable. At the same time, it is essential to recognize and connect with grassroots and community-driven initiatives. In Sri Lanka, for example, Yarl IT Hub provides free AI capacity-building courses, innovation challenges for youth, and mentorship opportunities. Such initiatives play a critical role in democratizing access to AI knowledge and nurturing local talent. The added value of the AI Dialogue lies in its ability to bridge global frameworks with regional strategies and local innovation. By connecting these layers, amplifying successful grassroots models, and enabling knowledge exchange, the Dialogue can ensure that AI governance is not only globally coordinated but also locally relevant, inclusive, and actionable.

Governments should play a central role in setting policy direction and enabling interoperability across national AI governance systems. However, their participation should go beyond negotiation to include sharing real-world regulatory experiences, especially from countries still developing AI frameworks. The private sector and technology companies can contribute by sharing technical expertise, risk mitigation practices, and responsible innovation models. Their role is critical in ensuring that safety, transparency, and accountability are embedded in AI system design from the start, rather than added later as compliance measures. Academia and research institutions can support the Dialogue through evidence-based analysis, ethical assessments, and impact studies, helping bridge the gap between high-level principles and practical implementation. Civil society organizations are essential for ensuring that human rights, inclusion, and accountability remain at the center of discussions, particularly representing marginalized and underrepresented communities. Youth groups and grassroots innovators also need structured space to contribute. For example, initiatives like Yarl IT Hub demonstrate how local ecosystems can build AI literacy, foster innovation, and empower young people—inputs that should inform global policy discussions. In terms of structure, the AI Dialogue should adopt a multi-layered and continuous format rather than a one-time conference. This could include: (1) thematic working groups aligned with key AI governance areas; (2) regional consultations to reflect local realities; and (3) a permanent multi-stakeholder advisory mechanism to ensure continuity between sessions. A digital platform for ongoing input, transparency, and knowledge sharing would further strengthen participation. Importantly, the Dialogue should ensure meaningful participation, not symbolic inclusion, with clear mechanisms to integrate stakeholder inputs into policy outcomes. Equal speaking time and funding for participation should be guaranteed. Remote interpretation, asynchronous contribution channels, and a dedicated "Capacity Fund" covering travel and connectivity for delegates from least developed countries are non-negotiable to prevent the Dialogue from replicating existing power imbalances. This structure would help transform the AI Dialogue into a living global governance process that is inclusive, responsive, and impact-driven.

Global AI discussions are currently dominated by a "seniority and gender bias," where panels are frequently composed of male experts from traditional academic or corporate backgrounds. This exclusion leaves out women and youth, who are often the primary drivers of grassroots innovation and the first to experience the societal impacts of AI deployment. In countries like Sri Lanka, failing to include these voices means that governance frameworks often overlook critical perspectives on social inclusion, ethical bias, and the long-term implications of AI on future generations. In additiona, there two critical communities remain underrepresented in global AI governance: rural and remote populations (including farmers, indigenous groups, and plantation workers) and local tech community organizers (like Yarl IT Hub in Sri Lanka who run grassroots capacity-building). Their absence means governance frameworks miss practical realities of how AI affects livelihoods, peacebuilding, and access. To fix this, the AI Dialogue must enforce inclusive representation mandates that move beyond tokenism. This requires a structural shift, such as implementing "diversity quotas" for speaking roles and creating dedicated platforms for youth-led organizations and women in tech.

1. Silent listening spaces—deliberate 30-minute blocks with no speaking, only curated audio testimonies from underrepresented communities played through headphones. This forces delegates to listen before responding, breaking habitual debate patterns. 2. Breakout session on Policy - The Dialogue should host sessions where countries can "test-drive" governance templates. Stakeholders can simulate how a human-rights-aligned AI policy would function in a resource-constrained or post-conflict environment, identifying gaps in monitoring or legal infrastructure before formal adoption.

1. Australia's AI Technical Standard for Government provides a comprehensive lifecycle approach covering three phases: "Discover" (ethical risk assessments, bias evaluation, human oversight), "Operate" (embedding AI into enterprise systems with safeguards), and "Retire" (controlled decommissioning). Complementing this, Australia launched GovAI, a collaboration platform and sandbox where public servants can test AI processes, tools, and training, enabling safe experimentation while maintaining public trust. 2. Grassroots initiatives like the Yarl IT Hub in Sri Lanka demonstrate a decentralized approach to governance. By providing free AI training and mentorship, these platforms ensure that "capacity-building" isn't just a top-down transfer of tech, but the creation of a local, skilled workforce that understands domestic cultural and linguistic nuances. This practice prevents "digital colonization" by ensuring local developers-including youth and women-are the ones building and auditing tools for their own communities.