CIRCANA- Senior Manager/ Secretary- Association of Nepali Origin

In my view, the first Global Dialogue on AI Governance would be a success if it leads to tangible actions that empower the Global South rather than simply imposing solutions. AI companies should take responsibility for compensating the climate impacts of their technologies on vulnerable regions and ensure that their investments and initiatives are done with local communities, not for them. This approach ensures ownership, relevance, and sustainable outcomes. Equally important is enabling countries in the Global South to leverage AI for their own development priorities. This includes knowledge transfer, capacity-building programs, and access to open-source AI tools, so these nations can develop solutions tailored to their unique social, economic, and environmental challenges. Finally, the dialogue should establish mechanisms to audit and mitigate biases in AI systems, particularly those that disproportionately affect marginalized populations. Transparent frameworks for accountability and ethical oversight will ensure AI serves the public good globally, rather than reinforcing inequalities. Success, therefore, would be measured not just by statements or agreements, but by clear commitments to climate compensation, collaborative capacity-building, and actionable frameworks for bias auditing—creating a more inclusive and equitable global AI ecosystem.

• Safe, secure and trustworthy AI
• AI capacity-building
• Open-source software, open data and open AI models
• Social, economic, ethical, cultural, linguistic and technical implications of AI

I selected these areas because they address both the opportunities and risks of AI while ensuring that its benefits are inclusive and globally equitable. Social, cultural, and linguistic implications: AI often struggles with native languages due to limited datasets. For example, when interacting with ChatGPT in Nepali, the output can feel robotic and emotionally flat. This demonstrates the need for collaboration between AI developers, governments, and the UN to collect diverse linguistic, literary, and cultural datasets, ensuring AI preserves cultural richness and emotional nuance. Safe, secure, and trustworthy AI: AI systems must be auditable and operate under clear, enforceable ethical frameworks. A common "AI constitution," facilitated by the UN, could set global standards for transparency, accountability, and human-centric safeguards-similar to the constitutional layer implemented in Claude. AI capacity-building: Equipping the Global South with infrastructure, AI literacy, and practical skills is crucial to closing the digital divide. Investments should focus on enabling local communities to develop and use AI for solutions tailored to their social and economic needs. Open-source software, open data, and open models: Transparency is essential for trust and innovation. Open-source AI software and datasets allow scrutiny of how models are trained, help prevent biases, and enable wider participation in AI research and development. Focusing on these pillars ensures that AI development is ethical, inclusive, and globally beneficial. By addressing technical, cultural, and governance challenges together, the Global Dialogue can set a foundation for AI that empowers all communities while mitigating harm.

Yes, one critical cross-cutting issue not fully captured by the listed themes is the need for climate impact compensation by AI companies, especially for the Global South. AI development and large-scale model training consume significant energy, contributing to carbon emissions that disproportionately affect vulnerable countries. A global framework should ensure that companies actively compensate for these environmental impacts through investments, technology transfer, and support for climate adaptation in affected regions.

In my country and region, governance gaps and limited AI capacity are creating both significant challenges and untapped opportunities. One major challenge is the rise of misinformation and deepfakes, which have already influenced public perception during recent elections. Without strong regulatory frameworks and public awareness, AI-driven disinformation can erode trust in institutions and social cohesion. Another critical issue is low AI literacy and limited adoption. Most users—including professionals—have little understanding of AI tools or their potential applications in daily work. Even when open-source AI models are available, there is a lack of knowledge on how to use them, which limits innovation and capacity-building. Finally, insufficient and biased datasets reduce AI's effectiveness for local contexts. For example, Nepali language datasets are sparse, causing AI outputs to be robotic and emotionally flat, and failing to capture the cultural richness and nuances of our society. This limits AI's usefulness in education, governance, and social services. However, these gaps also present opportunities. By developing AI literacy programs, building accessible infrastructure, and creating high-quality local datasets, we can empower users to leverage AI responsibly. Open-source AI initiatives, when combined with education and support, can enable communities to innovate and solve local challenges. Addressing these governance gaps—through regulation, capacity-building, and collaboration with international partners—can help ensure that AI benefits our society, preserves cultural identity, and mitigates risks such as disinformation and inequitable access.

The AI Dialogue can advance international cooperation by creating policies that include voices from the Global South, ensuring equitable representation in AI governance. It can facilitate capacity-building and AI literacy, empowering communities to understand and responsibly use AI. Additionally, it can help nations leverage AI for development, fostering collaboration, knowledge sharing, and innovation across borders while ensuring ethical and inclusive AI deployment globally.

The AI Dialogue should build upon existing initiatives like UNESCO's AI for Development programs, AI capacity-building partnerships in least developed and developing countries, and regional tech collaborations. By connecting with these efforts, it can amplify the voices of underrepresented nations, ensure inclusive policy-making, and provide targeted support for AI literacy, infrastructure, and ethical frameworks. Its added value lies in creating a coordinated global platform where developing countries can access resources, share best practices, and influence AI governance on equal footing with more developed states.

Different stakeholders can contribute to the AI Dialogue in complementary ways. Governments can propose policies and regulatory frameworks that ensure ethical, safe, and inclusive AI development. Private sector leaders and CEOs of major AI companies can share technological insights, best practices, and commit to transparency, accountability, and responsible AI deployment. Civil society, academia, and UN agencies can provide independent research, audit AI systems, and advocate for marginalized communities, including those in the Global South. For format and structure, the Dialogue could combine policy-focused plenaries, sector-specific working groups, and interactive workshops. Plenaries allow broad discussion on global standards, while working groups focus on thematic areas such as AI literacy, open-source initiatives, and climate impact compensation. Interactive workshops can foster practical collaboration, capacity-building, and actionable commitments. A rotating leadership model with representation from developing countries can ensure equitable participation and sustained impact.

Voices from youth in the Global South, like myself, are often underrepresented in AI governance discussions. They can be included through dedicated youth panels, mentorship programs, and participatory workshops that allow young people from developing countries to contribute insights, co-develop policies, and influence AI deployment priorities. Partnering with local universities, civil society organizations, and UN agencies can ensure meaningful engagement and amplify these perspectives in global forums.

Meaningful engagement in the AI Dialogue can be achieved through interactive, hands-on formats. Live demonstrations of large language models (LLMs) can show how AI works and where biases appear. Auditing workshops allow participants to examine outputs for fairness, especially in underrepresented languages. Open-source showcases let attendees explore, adapt, and learn from transparent AI tools. Cross-sector working groups or hackathons can address issues like climate impact compensation, capacity-building, and equitable AI use. Combining these approaches with digital platforms ensures youth and voices from the Global South can actively participate, making the Dialogue both informative and action-oriented.

Several policies and practices can promote effective AI governance and address its challenges. One key approach is climate impact compensation: AI companies should assess and offset the carbon footprint of their technologies, particularly when their energy-intensive operations affect vulnerable countries in the Global South. Such policies ensure that AI development does not exacerbate environmental inequalities. Another essential practice is bias auditing. Regular independent audits of AI systems can identify and mitigate social, cultural, and linguistic biases, ensuring fairness and inclusivity. This is particularly important for underrepresented languages and communities, which are often excluded from training datasets. Some AI platforms, like Claude, have implemented a constitutional layer, which defines ethical rules and operational boundaries for the model. Policies that require a similar framework globally can ensure AI systems follow consistent ethical and human-centric standards. Partnerships and collaboration also strengthen governance. Governments, UN agencies, academia, civil society, and private companies can co-develop AI policies, share datasets, and implement capacity-building initiatives. Partnerships ensure that solutions are contextually relevant, culturally sensitive, and accessible to the Global South. Additionally, open-source models and data platforms promote transparency and accountability, allowing stakeholders to understand how AI is trained, test its outputs, and adapt technologies for local needs.