Secretariat of the Minamata Convention on Mercury (UNEP)

A successful first Global Dialogue on AI Governance would be one that translates shared principles into practical, actionable pathways for the responsible use of AI in support of multilateral processes. In particular, it would be valuable to advance a common understanding of how AI can be applied in data-rich policy environments, such as multilateral environmental agreements, while ensuring trust, transparency and accountability. For the Minamata Convention on Mercury and other MEAs, AI holds particular promise in multilingual translation, and transforming data like national reporting or COP decisions into actionable insights, thereby supporting evidence-based decision-making and more effective information exchange. The Dialogue could play an important role in advancing this shift from data availability to knowledge-driven implementation.

• AI capacity-building
• Interoperability of governance approaches
• Protection and promotion of human rights
• Transparency, accountability, and human oversight

Capacity-building remains essential to ensure that all countries, in particular developing countries, can effectively engage with and benefit from AI applications. In the context of multilateral environmental agreements, this is closely linked to strengthening national reporting systems and the use of data for implementation. Interoperability is equally critical. Multilateral processes rely on multiple data systems and governance frameworks, and there is a growing need for harmonized standards, shared taxonomies and compatible digital infrastructures to enable meaningful data exchange and analysis. Transparency, accountability and human oversight are central to maintaining trust, particularly where AI supports the analysis of official data and informs policy processes. Ensuring that outputs remain verifiable and subject to expert validation is key. The protection and promotion of human rights remains a foundational pillar of the United Nations system and should continue to guide the development and use of artificial intelligence.

First, the environmental sustainability of AI. As AI systems expand, their energy consumption, resource use and environmental footprint are becoming increasingly significant. It is important to ensure that AI governance frameworks consider these impacts and promote approaches that are aligned with environmental sustainability objectives, including within the context of global environmental agreements. Second, data sovereignty and equitable access to data. There is a growing need to avoid reinforcing existing asymmetries in access to data and digital infrastructure, which could lead to new forms of dependency or "digital colonialism". Ensuring that countries retain meaningful control over their data, while enabling fair and inclusive data sharing, will be critical for building trust and supporting balanced participation in AI development and governance.

In the context of multilateral environmental agreements, governance gaps are most evident in the limited capacity to fully leverage large volumes of narrative and decentralized data, such as national reports, scientific data, and project information managed across different institutions. Key challenges include the lack of standardized data structures and interoperability across systems, particularly in relation to projects implemented by external partners. This affects the ability to generate coherent, comparable insights. In addition, it is essential to ensure that the development and use of AI does not inadvertently exclude important perspectives, including those of civil society, Indigenous Peoples and local communities. At the same time, there are significant opportunities. AI can enhance transparency and traceability, including by supporting better tracking of decisions adopted by Conferences of the Parties and their implementation over time. It can also improve the analysis of national reporting data, enabling the identification of trends, challenges and good practices. Within the Minamata Convention, exploratory work is underway to apply AI to the analysis of national reports. This has the potential to strengthen information exchange and contribute to a more effective implementation of the Convention, transforming reporting processes into dynamic knowledge systems that better support Parties.

The Dialogue has the potential to move the international community from fragmented experimentation towards more coherent and coordinated approaches to AI governance. It can facilitate the exchange of concrete experiences and use cases from across sectors, including the United Nations system and multilateral environmental agreements, and help identify common approaches to responsible use of AI. The Dialogue can also contribute to advancing shared standards and principles, particularly in relation to data governance, interoperability and accountability, while supporting capacity-building efforts tailored to public sector needs.

There are several ongoing initiatives that provide a strong foundation for the AI Dialogue to build upon. These include system-wide efforts such as the UN 2.0 agenda, the Global Digital Compact, and emerging guidance on the responsible use of AI within the United Nations. Within the environmental domain, platforms such as InforMEA and the digital ecosystems developed by multilateral environmental agreements (MEAs), including the Minamata Convention, are advancing work on data interoperability, knowledge management and digital transformation. The Convention's digital strategy aims to support structured and accessible information exchange among Parties. In addition, partnerships with networks such as the Geneva Environment Network contribute to strengthening outreach, collaboration and knowledge-sharing across the environmental community.

Inclusive participation could be supported through a combination of policy dialogue and practical engagement formats. This may include high-level discussions complemented by technical sessions and case-based exchanges, allowing stakeholders to share concrete experiences and lessons learned. Ensuring accessibility will be key, including through hybrid participation, multilingual formats and open consultation mechanisms. Engagement should extend to governments, International Organisations, academia, civil society and the private sector. Equally important is that the outputs of the Dialogue are practical, accessible and reusable, enabling stakeholders to apply the insights generated in their respective contexts.

Certain perspectives remain underrepresented in global discussions on AI governance, including those of developing countries with limited digital infrastructure. In the environmental context, national focal points, technical experts, and local actors, including Indigenous Peoples and local communities, bring valuable knowledge that is often not sufficiently reflected. Their inclusion could be strengthened through targeted capacity-building, dedicated consultation mechanisms, and efforts to ensure accessibility, including multilingual and low-bandwidth participation options. Recognizing and integrating these perspectives would contribute to more inclusive and context-sensitive approaches to AI governance.

Innovative formats could help ensure that the Dialogue remains both practical and engaging. These may include use-case driven sessions, where participants explore concrete applications of AI in areas such as data analysis or knowledge management, as well as interactive demonstrations of tools. Problem-solving workshops could also be valuable, bringing together diverse stakeholders to address specific governance challenges. Cross-sector roundtables would further support dialogue between policy, technical and operational perspectives.

The work of the Minamata Convention offers an example of how AI can be integrated into knowledge management and implementation processes in a responsible and targeted manner. The Secretariat of the Convention has developed a digital ecosystem that includes an online reporting tool, data dashboards and plans for a centralized Exchange Platform to facilitate structured information exchange. Building on this foundation, exploratory work is being undertaken to apply AI to the analysis of national reports, with the aim of identifying trends, challenges and good practices. This approach is guided by key principles, including the use of official and validated data sources, human oversight in the validation of outputs, and the development of metadata and taxonomies to support reproducibility and interoperability. It is also aligned with broader UN guidance on the responsible use of AI, including data protection and accountability considerations.