Indigenous Data Network - University of Melbourne

The first Global Dialogue on AI Governance would be successful if it establishes a shared and practical foundation for international cooperation that can guide national, regional and sectoral approaches over time. Rather than limiting itself to aspirational principles, the Dialogue should deliver outcomes that clarify how AI governance can be implemented across diverse legal, cultural and institutional contexts. Success would involve clear articulation of baseline global expectations for AI governance that are grounded in international human rights law and attentive to both individual and collective rights. Central to this is explicit recognition that data governance underpins AI governance, including governance across the full data lifecycle from initial collection and generation through to reuse, linkage and retention. A successful Dialogue would also contribute to agreement on a minimum set of governance functions, such as data stewardship, risk assessment, transparency mechanisms, accountability pathways and avenues for redress, while allowing flexibility in regulatory form. Commitments to interoperability of governance approaches are important in this context, enabling cross-border operation of AI systems and data flows without undermining local governance authority. Equally important is the establishment of credible pathways for capacity-building. Many communities, institutions and countries face structural barriers to participating meaningfully in AI governance, and the Dialogue should support capability development alongside norm-setting. Experience from Indigenous‑led national data infrastructure initiatives coordinated through the Indigenous Data Network demonstrates that governance capability and community‑controlled data infrastructure are preconditions for effective participation in data‑intensive systems. Finally, mechanisms such as standing working groups, review cycles or follow-on processes would help translate dialogue into implementation and adaptation over time. These outcomes would support the emergence of a globally recognisable benchmark for AI governance that is technically credible, normatively grounded and responsive to existing inequalities in data and digital systems.

• Protection and promotion of human rights
• Safe, secure and trustworthy AI
• AI capacity-building
• Social, economic, ethical, cultural, linguistic and technical implications of AI

The selected priorities reflect areas where urgent global action is required to prevent AI systems from entrenching existing inequalities, particularly those arising from historical and ongoing extractive data practices affecting Indigenous peoples. The protection and promotion of human rights is fundamental because AI governance must address not only individual rights but also collective rights, including rights to culture, self-determination and authority over data relating to peoples, lands, languages and knowledge systems. For Indigenous Peoples, these rights are grounded in international law and long-standing governance systems. Without explicit recognition of collective rights and governance authority, AI governance frameworks risk normalising the use of Indigenous data without lawful authority, consent or accountability. This risk has been observed in practice across multiple sectors addressed through Indigenous Data Network-supported governance work. Safe, secure and trustworthy AI is a priority because trustworthiness depends on more than technical performance. It requires attention to data provenance, lawful data access, appropriate use, and mechanisms to prevent cultural, social and community-level harm. As AI systems increasingly retrieve and act on data dynamically, governance failures can occur at scale if questions of authority and accountability are not addressed. AI capacity-building is essential to enable meaningful participation in AI governance. Many Indigenous communities and institutions face structural barriers, including limited access to technical infrastructure, governance resourcing and institutional support. Capacity-building must therefore encompass governance capability, community-controlled data infrastructures and the ability to make informed decisions about if, how and under what conditions AI systems engage with Indigenous data. Finally, the social, economic, ethical, cultural, linguistic and technical implications of AI require focused attention because AI systems directly affect Indigenous languages, knowledge transmission, representation and service delivery. Understanding these impacts is necessary to ensure AI contributes to equitable and culturally grounded development rather than reinforcing historical patterns of extraction and misrepresentation.

Several cross-cutting and emerging issues require more explicit attention within the Global Dialogue, as they underpin the effectiveness and legitimacy of AI governance across all thematic areas. First, the governance of increasingly agentic and autonomous AI systems remains insufficiently addressed. These systems can generate, retrieve, combine and act on data with limited direct human instruction, blurring established distinctions between training data, operational data and AI-generated data. This raises unresolved questions about authority, accountability, consent and responsibility across the data lifecycle, particularly where systems operate across jurisdictions and governance regimes. The foundational relationship between AI governance and data governance also requires stronger and more explicit recognition. AI governance frameworks often assume that data inputs are neutral or lawfully available, yet many AI-related harms arise from weak or absent data governance. Issues of provenance, stewardship, reuse, linkage and retention are central to determining whether AI systems are lawful, trustworthy and socially acceptable. Treating data governance as implicit risks undermining AI governance objectives. Further, the recognition of Indigenous Peoples as governance authorities remains a critical gap. Indigenous data is frequently incorporated into AI systems without appropriate authority, consent or accountability, despite the existence of collective rights grounded in international law and long-standing Indigenous governance systems. These issues cannot be adequately addressed through generic ethics or inclusion frameworks, as they concern jurisdiction, decision-making authority and collective rights. Explicit recognition of Indigenous governance frameworks is therefore a cross-cutting governance requirement. Finally, limited attention is given to the political economy of AI, including the concentration of data, computational resources and decision-making power. Without addressing these structural asymmetries, global AI governance risks reinforcing existing inequalities rather than mitigating them. Addressing these cross-cutting issues would strengthen the coherence, fairness and future readiness of the Global Dialogue's outcomes.

Governance gaps in AI and data governance are already producing tangible effects across multiple sectors, with particularly pronounced consequences for Indigenous Peoples and institutions. In contexts such as health, education, social services and cultural heritage, AI systems are increasingly deployed using data that may be incomplete, misclassified or collected without appropriate authority. Where collective rights and Indigenous governance frameworks are not recognised, these systems risk causing cultural harm, reinforcing deficit narratives and eroding trust. One of the most significant challenges arises from weak governance of data reuse, linkage and secondary analysis. Even where initial data collection may have occurred within a governed context, subsequent uses for AI development or deployment often proceed without renewed consent or accountability. As AI systems become more autonomous and capable of operating across dynamic data environments, these governance gaps can scale rapidly, with limited avenues for redress. Capacity constraints further exacerbate these risks as many Indigenous communities and organisations face structural barriers to engaging with AI governance, including limited access to technical infrastructure, governance resourcing and specialist expertise. Indigenous Data Network engagement across research, health and administrative data contexts has shown that these constraints materially limit Indigenous participation in AI‑related decision‑making. However, there are also emerging opportunities. Increased global focus on AI governance has created space to strengthen data governance arrangements, embed Indigenous governance frameworks and invest in community-controlled data infrastructure. When coupled with genuine capacity-building, these developments offer pathways to more equitable participation and governance arrangements that support Indigenous aspirations. Addressing current governance gaps is therefore necessary both to mitigate harm and to enable AI systems to operate in ways that are lawful, accountable and socially legitimate.

The AI Dialogue can advance international cooperation on AI governance by providing a globally legitimate forum capable of aligning diverse governance systems while respecting different legal, cultural and institutional authorities. A key role is to support convergence around shared baseline expectations for AI governance without imposing uniform regulatory models. This is particularly important for AI systems and data flows that operate across borders, where fragmented governance arrangements can undermine accountability, trust and legal clarity. By identifying minimum governance functions and points of interoperability, the Dialogue can help reduce fragmentation while preserving local decision-making authority. The Dialogue can also facilitate cooperation through sustained knowledge exchange and capacity-building. Many countries, communities and institutions face structural barriers to implementing even basic AI governance measures. A global platform that shares governance practices, institutional models and lessons learned can help build capability in ways that are responsive to context rather than prescriptive. Importantly, the Dialogue has a critical role in advancing recognition of Indigenous Peoples as governance authorities within international AI governance. Cooperation that fails to account for collective rights, consent frameworks and Indigenous governance systems risks reproducing extractive data practices at a global scale. By creating space for Indigenous participation as decision-makers and knowledge holders, the Dialogue can strengthen the legitimacy and effectiveness of international cooperation. Finally, as AI-related risks increasingly transcend national boundaries, the Dialogue can support cooperation on risk identification, harm mitigation and accountability, particularly for more autonomous and agentic systems. Through continuity mechanisms and follow-on processes, the Dialogue can help translate cooperation into sustained governance practice. The AI Dialogue can function not only as a forum for discussion but as an enabling mechanism for coordinated, rights-based and globally relevant AI governance.

The AI Dialogue has the opportunity to add coherence and legitimacy to a landscape of existing initiatives by connecting, rather than duplicating, current global governance efforts. Work undertaken through the UN system on data governance provides a critical foundation. In particular, outcomes from the UN Commission on Science and Technology for Development Working Group on Data Governance offer negotiated guidance on data stewardship, equity, benefit-sharing and trusted data flows. Given that AI systems are dependent on data across their full lifecycle, stronger alignment between AI governance discussions and existing data governance frameworks would significantly strengthen global coherence. International normative instruments such as UNESCO's Recommendation on the Ethics of AI and the OECD AI Principles also provide widely recognised reference points. Their value lies in their adaptability across jurisdictions and their emphasis on rights-based governance. However, these instruments often under-specify how collective rights, data governance authority and culturally grounded governance arrangements should be operationalised. Technical standards bodies contribute practical mechanisms for implementing governance principles in areas such as risk management, transparency and data quality. These tools are important, but insufficient to address questions of lawful authority, consent and accountability, especially when data relates to Indigenous people and knowledge systems. The added value of the AI Dialogue lies in its capacity to integrate these initiatives within a broader governance architecture that explicitly recognises diverse governance authorities. By connecting AI governance with Indigenous data governance frameworks, including community-controlled data infrastructures and consent-based governance models, the Dialogue can address persistent gaps between normative commitments and on-the-ground practice. The Dialogue can help shift global AI governance toward approaches that are technically sound and also socially legitimate, legally grounded and responsive to longstanding inequities in data systems.

Effective contribution to the AI Dialogue requires formats and structures that recognise different forms of authority, expertise and resourcing, rather than assuming a single model of participation. Governments, technical experts, industry, civil society, Indigenous Peoples and community institutions all bring distinct knowledge and governance responsibilities that should shape how engagement is structured. The Dialogue should therefore enable multiple modes of participation, combining in‑person meetings with virtual and asynchronous mechanisms. This is necessary to address geographic, financial and institutional barriers that limit participation by many communities and organisations. Participation should not depend solely on the ability to attend international meetings in person. A structured, layered approach would help ensure both breadth and depth of engagement. This could include open consultations for broad input, thematic working groups focused on specific governance challenges, and regional dialogues that surface context‑specific priorities and governance arrangements. For Indigenous Peoples, dedicated pathways are required that respect existing governance protocols and decision‑making processes, rather than subsuming participation within general civil society formats. Stakeholder contribution should also be continuous rather than episodic. Mechanisms such as iterative drafting, transparent publication of submissions and opportunities for feedback on evolving proposals would strengthen accountability and trust. Importantly, the Dialogue should be oriented toward co‑development of governance approaches, rather than limiting stakeholder roles to consultation or commentary.

Indigenous Peoples remain significantly underrepresented in global AI governance discussions, despite being deeply affected by data‑driven systems and digital transformation. This underrepresentation reflects a broader tendency to frame Indigenous Peoples as stakeholders or beneficiaries rather than as governance authorities with distinct rights, jurisdictions and knowledge systems. The consequences of this exclusion are substantive. AI systems are frequently developed and deployed using data relating to Indigenous peoples, lands, languages and knowledge systems, often without appropriate authority or accountability. In the absence of Indigenous governance frameworks, these systems can perpetuate misrepresentation, cultural harm and extractive data practices, even where they are framed as ethically motivated or socially beneficial. Meaningful inclusion requires structural change rather than expanded consultation alone. Indigenous Peoples should be recognised as governance authorities in AI and data governance processes, with dedicated roles in decision‑making bodies, thematic working groups and advisory mechanisms. Consultation pathways should be designed to align with Indigenous governance structures and cultural protocols, including collective decision‑making and consent processes. Practical measures are also required to support participation. These include resourcing for sustained engagement, translation and accessibility support, and investment in Indigenous‑led institutions and expertise. Embedding Indigenous data governance frameworks into global AI governance benchmarks would further signal that Indigenous perspectives are integral to governance, not ancillary. Addressing this underrepresentation is essential both for the protection of rights and for the legitimacy of global AI governance outcomes.

Meaningful engagement in the AI Dialogue will require formats that move beyond conventional plenary sessions and written submissions, toward approaches that support dialogue, co‑development and cultural legitimacy. Hybrid participation models should be standard practice, combining in‑person meetings with virtual and asynchronous engagement. Low‑bandwidth and time‑zone‑sensitive options are essential to enable participation by communities and institutions with limited digital infrastructure or resourcing. Regional dialogues can further support engagement by grounding discussions in local governance contexts and feeding insights into the global process. Culturally grounded engagement formats are particularly important. Indigenous‑led dialogues, community roundtables and other culturally appropriate forums can support forms of discussion and decision‑making that differ from standard international conference models, while remaining fully relevant to global governance outcomes. These formats help ensure that Indigenous knowledge systems and governance principles inform, rather than merely respond to, AI governance discussions. The Dialogue could also benefit from facilitated co‑creation environments, where policymakers, technical experts, community representatives and rights‑holders work together to develop specific governance tools, benchmarks or safeguards. These settings should be iterative and problem‑focused, with clear pathways for integrating outputs into the formal Dialogue process. Finally, transparent and continuous engagement platforms, including public drafts and feedback loops, can help sustain participation over time and strengthen accountability. Innovative formats should serve not novelty, but the deeper goal of legitimacy, inclusiveness and durable governance outcomes.

Effective AI governance is most credible where it builds on strong data governance practices and recognises lawful authority over data across its full lifecycle. Approaches emerging from global work on data governance provide important foundations for addressing AI-related risks in practical ways. Frameworks developed through the United Nations system on data governance illustrate how principles such as stewardship, equity, benefit-sharing and trusted data flows can be translated into governance arrangements that shape how data is collected, accessed, reused and linked. These approaches are particularly relevant to AI systems, which depend on data not only at the point of model development but throughout deployment and adaptation. Indigenous data governance frameworks offer concrete and well-established practices that address governance challenges often left unresolved in AI discussions. These include collective decision-making over data use, consent-based governance arrangements, community-defined conditions of access and use, and accountability mechanisms grounded in long-standing governance systems. Where implemented through community-controlled data infrastructures and institutions, these approaches demonstrate how data-intensive systems can operate in ways that respect collective rights and cultural integrity. International normative instruments on AI ethics and human rights provide additional guidance, particularly where they emphasise accountability, human oversight and proportional risk management. Technical standards developed by international standards bodies further support implementation by offering tools for assessing risk, transparency and data quality. However, these instruments are most effective when applied within governance frameworks that address questions of authority, consent and responsibility. Taken together, these practices show that effective AI governance is not reliant on a single model or tool. Rather, it depends on integrating technical standards, rights-based frameworks and data governance approaches that are responsive to social and cultural context. Strengthening these linkages is essential for AI governance that is both operational and legitimate.