Bangladesh NGOs Network for Radio and Communication

In my view, the first Global Dialogue on AI Governance would be successful if it delivers an outcome that is inclusive, actionable and responsive to the priorities of developing regions, including South Asia. The Global Dialogue has been established as an inclusive United Nations platform on AI governance, and its inaugural session is scheduled to take place in Geneva on 6–7 July 2026, alongside the AI for Good Global Summit. From a South Asian perspective, success should be measured not only by the quality of the discussion, but by whether it helps build shared understanding, practical cooperation and equitable implementation. First, the Dialogue should advance a common foundation for AI governance rooted in human rights, safety, transparency, accountability and public interest. At the same time, it should respect different national contexts and development realities. Second, it should place strong emphasis on capacity-building and AI equity. For South Asia, this means support for skills development, institutional readiness, affordable access to computing resources, local-language data and context-sensitive innovation. Without these elements, AI governance risks reinforcing existing inequalities rather than addressing them. This concern is particularly relevant in countries where digital divides remain significant. Third, success would require meaningful participation by the Global South. Governments, civil society, academia, technical communities, women, youth and local innovators from South Asia should not only be consulted, but should help shape priorities and solutions. Ultimately, the Dialogue will be successful if it moves beyond broad principles and helps create a more inclusive global AI governance ecosystem, one that enables South Asia and other developing regions to participate as equal partners in shaping a safe, just and development-oriented AI future.

• Safe, secure and trustworthy AI
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• AI capacity-building
• Transparency, accountability, and human oversight

From our perspective, all four thematic areas identified in General Assembly resolution 79/325 are important, but the most urgent priorities for our entity are AI capacity-building; transparency, accountability and human oversight; and the social, economic, ethical, cultural, linguistic and technical implications of AI. These thematic areas are central to the Global Dialogue's mandate and are especially relevant for developing-country contexts. AI capacity-building is a priority because many countries in the Global South still face major gaps in skills, institutional readiness, data governance, computing access and policy expertise. Without meaningful capacity-building, participation in AI governance will remain uneven, and the benefits of AI will be concentrated in a few countries and companies. Transparency, accountability and human oversight are equally urgent. For our entity, these are essential to safeguarding human rights, public trust, information integrity and democratic participation. Human oversight is particularly important where AI systems influence access to information, media environments, public services or decision-making that affects people's lives. We also attach high importance to the social, economic, ethical, cultural, linguistic and technical implications of AI. In multilingual and socially diverse contexts such as South Asia, AI governance must address exclusion, bias and unequal representation. AI systems should respect local languages, cultural diversity and the realities of marginalized communities. Finally, safe, secure and trustworthy AI remains a cross-cutting foundation. In our view, safety must go hand in hand with equity, inclusion and meaningful participation by developing countries and civil society if AI governance is to be credible, effective and development-oriented.

Yes. While the listed themes are highly relevant, in our view there are several cross-cutting and emerging issues that would merit more explicit attention in the first Global Dialogue on AI Governance. As the Dialogue is intended to serve as an inclusive global platform on AI governance, it would benefit from recognising issues that cut across safety, capacity, societal impact and accountability. First, information integrity and democratic resilience deserve explicit focus. AI is already affecting public discourse, electoral environments, journalism and civic trust through synthetic media, automated influence operations and large-scale manipulation. This issue is broader than safety alone and has direct implications for human rights and democratic governance. Second, concentration of data, compute and market power should be addressed more clearly. AI governance is not only about how systems behave, but also about who controls the infrastructure, models, datasets and platforms. For developing countries, this concentration may deepen dependency and widen existing digital inequalities. Third, environmental sustainability and resource justice should be treated as a distinct concern. The energy, water and material demands of AI systems have implications for climate resilience and sustainable development, especially in countries already facing resource constraints. Fourth, access to remedy and redress should receive greater attention. People and communities affected by harmful AI-enabled decisions need practical, affordable and rights-based mechanisms to seek explanation, correction and remedy. Finally, linguistic and cultural equity in practice should remain a priority. For regions such as South Asia, meaningful inclusion depends not only on language recognition, but on whether AI systems reflect local knowledge, diverse contexts and the realities of underrepresented communities. Addressing these issues explicitly would help ensure that the Dialogue is more balanced, development-oriented and responsive to the needs of the Global South.

In Bangladesh, and across South Asia, the most significant challenge is that AI development is advancing faster than governance readiness. While policy attention is growing, important gaps remain in implementation capacity, institutional coordination, transparency, public oversight and redress. In Bangladesh, UNESCO's national readiness assessment notes that the draft National AI Policy is still under review and that concrete implementation steps, public communication mechanisms and transparency arrangements remain to be established. This points to a broader regional challenge: many institutions are still building the legal, technical and human capacities needed to govern AI in a rights-based and development-oriented manner. A second major challenge concerns inclusion and inequality. Across South Asia, unequal connectivity, uneven access to compute, limited local-language data, and gaps in digital skills risk deepening existing social and economic divides. The World Bank underscores the importance of the "four Cs" of AI readiness: connectivity, compute, context and competency. At the same time, UNDP warns that poorly governed AI can heighten exclusion, enable surveillance, weaken accountability and produce "algorithmic exclusion" when datasets fail to reflect marginalized communities. These risks are particularly relevant in diverse and multilingual societies such as those of South Asia. At the same time, the opportunities are substantial. If supported by inclusive governance, AI can strengthen public service delivery, agriculture, healthcare, education, disaster preparedness and financial inclusion. For Bangladesh and South Asia, the most promising pathway is therefore not unregulated acceleration, but responsible and inclusive adoption grounded in capacity-building, local-language innovation, representative data, and effective accountability mechanisms

In our view, the Global Dialogue on AI Governance can play a valuable role by serving as a trusted multilateral space for convergence. The United Nations has framed the Dialogue as an inclusive platform for States and stakeholders to discuss critical AI governance issues, while related UN materials emphasize its function in sharing practices, building common understanding and fostering interoperability across approaches. First, the Dialogue can help reduce fragmentation in the global AI governance landscape. Today, countries and regions are developing rules, standards and policy frameworks at different speeds and with different priorities. The Dialogue can support greater coherence by identifying common principles and practical areas of alignment, while respecting diverse national contexts and development needs. This is particularly important for developing countries, which often face the burden of navigating multiple, overlapping governance models. Second, it can strengthen international cooperation through peer learning and capacity-building. By bringing together governments, civil society, academia, the private sector and technical communities, the Dialogue can help share experiences on implementation, risk governance, institutional readiness and public-interest safeguards. For countries in the Global South, this can support more meaningful participation in shaping AI governance rather than remaining primarily rule-takers. Third, the Dialogue can help ensure that AI governance remains inclusive, development-oriented and anchored in the UN system. The broader UN vision is to lay the foundations for a globally inclusive and distributed architecture for AI governance based on international cooperation. In that sense, the Dialogue can become a bridge between global norms and local realities, helping translate broad principles into more equitable and actionable cooperation. Ultimately, its value will lie in helping the international community move from parallel discussions toward more coordinated, inclusive and implementation-focused cooperation on AI governance.

In the Bangladesh and South Asia context, the AI Dialogue should build on at least four existing foundations. First, the UNESCO Recommendation on the Ethics of AI and its implementation tools, especially the Readiness Assessment Methodology (RAM), provide a practical basis for translating high-level principles into national action. Bangladesh has already completed the RAM process, which offers a useful evidence base on governance readiness, institutional gaps and country priorities. Second, the Dialogue should connect with the Internet Governance Forum (IGF), which the United Nations has already identified as a complementary forum for AI-related policy discussion. Third, the Dialogue should link closely with AI for Good, which is organized by ITU with more than 50 UN entities and already brings together work on standards, skills, solutions and policy exchanges. Fourth, from a South Asian perspective, it should also connect with relevant regional and national efforts such as India's IndiaAI Mission and Global IndiaAI Summit, which have focused on compute capacity, datasets, skills, startups and safe and trusted AI, while also engaging GPAI processes. These platforms show that the region is not only a recipient of AI policy, but also an emerging contributor to it. The added value of the AI Dialogue would be to connect these currently fragmented efforts into a more coherent multilateral ecosystem. It can help align ethics, capacity-building, standards and development priorities across institutions, while ensuring that the needs of Bangladesh, South Asia and the wider Global South are better reflected in global discussions. In particular, the Dialogue can add value by elevating issues such as local-language AI, public-sector readiness, inclusive datasets, and equitable access to compute and knowledge. In this way, it can serve not as a duplicate platform, but as a bridge between existing initiatives and more coordinated international cooperation.

Different stakeholders can contribute to the AI Dialogue in distinct but complementary ways. Governments can share policy experience, regulatory lessons and public-interest priorities. Private sector actors can contribute technical expertise, risk information and implementation experience. Civil society can bring human rights, inclusion and accountability perspectives. Technical and academic communities can provide independent evidence, standards-related insight and foresight on emerging risks. Youth, women's groups, workers' representatives, persons with disabilities, Indigenous Peoples and local communities can help ensure that the Dialogue reflects lived realities and does not remain confined to institutional viewpoints. The United Nations has framed the Global Dialogue on AI Governance as an inclusive platform within the UN for States and stakeholders, which makes such participation essential. In terms of format and structure, we would recommend a layered and participatory model. The current concept note already points toward a useful foundation through a multistakeholder plenary meeting, a high-level governmental segment and thematic discussions. This should be complemented by: Open written submissions and regional consultations in advance, to capture diverse views, especially from developing countries. Balanced thematic roundtables with moderated participation from all stakeholder groups, not parallel silos. Dedicated sessions for underrepresented voices, including youth, local-language communities and Global South institutions. Hybrid and multilingual participation, with remote access, interpretation and accessible documentation. A concise Chair's summary with follow-up pathways, identifying areas of convergence, open questions and opportunities for cooperation. The added value of this approach would be to make the Dialogue not only inclusive in principle, but also inclusive in practice. It would help connect technical knowledge, policy experience and community perspectives, while building trust and broad ownership of the AI governance agenda.

In the Bangladesh and South Asia context, several voices remain underrepresented in global AI governance discussions: rural and low-income communities, women and girls, persons with disabilities, minority and local-language communities, small innovators and SMEs, and public-interest civil society actors working closest to affected communities. This matters because the region's digital divides are still significant. In Bangladesh, only about 44.5 per cent of the population uses the Internet, and UNESCO has highlighted the need for greater attention to AI systems in Bengali and minority ethnic languages, as well as more diversified datasets. At the regional level, the World Bank has stressed that digital policies in South Asia must be designed to be accessible especially for women, persons with disabilities and minorities, while also equipping SMEs with the tools to participate. A second gap concerns representation from the Global South in agenda-setting itself. UN reporting has warned that AI expertise and decision-making remain concentrated in a small number of countries and companies. As a result, many communities in South Asia are more often policy takers than policy shapers, even though they are deeply affected by AI in areas such as public services, labour, language technologies, education and information integrity. To address this, the AI Dialogue should include regional consultations in South Asia, multilingual participation, open written submissions, hybrid access, travel support for underrepresented participants, and dedicated space for community-based and local-language evidence. It should also ensure that women's groups, disability advocates, youth, local researchers, minority-language experts and grassroots digital rights actors are not only consulted, but meaningfully represented in thematic sessions and follow-up processes. The UN has already framed the Dialogue as an inclusive platform and related consultation inputs have pointed to the value of regional forums to ensure diverse representation.

To foster meaningful and dynamic engagement, the AI Dialogue should combine formal multilateral exchange with interactive, problem-solving formats. This would be consistent with the UN's vision of the Global Dialogue as an inclusive platform involving Governments and all relevant stakeholders, and with the current concept note, which already foresees a multistakeholder plenary meeting, a high-level governmental segment and thematic discussions. In our view, the most effective innovative formats would include moderated roundtables, scenario-based policy labs, regional listening sessions, and multi-stakeholder "issue clinics." Roundtables can support balanced participation across stakeholder groups. Scenario-based policy labs can test practical responses to concrete challenges such as AI in elections, education, health, public services or disaster response. Regional listening sessions, including for South Asia and other underrepresented regions, can bring local realities into the global process. Issue clinics could bring together governments, technical experts, civil society and industry to jointly examine one governance problem and identify possible cooperative responses. The Dialogue would also benefit from reverse panels, where communities most affected by AI systems speak first and institutions respond afterwards; fishbowl discussions, which allow dynamic participation beyond fixed speakers; and short evidence briefings from academia and technical experts before negotiations or roundtables begin. These formats would help move the discussion beyond prepared statements toward exchange, learning and practical convergence. To make these formats effective, they should be supported by hybrid participation, multilingual access, written input channels, rapporteur summaries and clear follow-up pathways. Existing UN discussions have already emphasized the need for remote participation, capacity support and more effective inclusion of stakeholders from developing countries and underrepresented groups. Ultimately, the added value of such formats is that they can make the AI Dialogue not only inclusive in principle, but also participatory, evidence-informed and solution-oriented in practice.

Yes. Several existing policies, practices, platforms and approaches already offer useful models for effective AI governance, including for Bangladesh and South Asia. First, the UNESCO Recommendation on the Ethics of AI and its Readiness Assessment Methodology (RAM) provide a practical, rights-based approach for translating broad principles into national action. For Bangladesh, this is especially relevant because the country has already completed the RAM process, which can help identify governance gaps, institutional priorities and implementation pathways. Second, the OECD AI Principles and the OECD AI Policy Observatory offer a strong example of how shared principles can be combined with evidence, policy tracking and peer learning. This approach is valuable because it supports both trustworthy AI and continuous policy learning, rather than treating governance as a one-time regulatory exercise. Third, the European Union AI Act is an important example of a risk-based regulatory framework. Its added value lies in differentiating between levels of risk and applying stronger safeguards where AI systems can significantly affect rights, safety or public trust. This type of approach may offer useful lessons for countries in South Asia as they develop context-specific frameworks. Fourth, in the regional context, IndiaAI Mission, including its Safe & Trusted AI pillar and the IndiaAI Safety Institute, provides an example of combining innovation support with governance measures. This is particularly relevant for South Asia because it links safety, inclusion, skills, datasets and ecosystem-building rather than treating governance and innovation as separate agendas. Taken together, these examples suggest that effective AI governance works best when it combines ethical principles, readiness assessments, risk-based regulation, evidence platforms, and capacity-building for local implementation. For Bangladesh and South Asia, this integrated approach may be more useful than adopting isolated rules alone