Mouvman Lavwa Dime

Success cannot be measured by the production of another non-binding communiqué. The first Global Dialogue will be a genuine success only if it generates three concrete, verifiable outcomes. First, a negotiation mandate. The Dialogue convenes four months before the CCW Seventh Review Conference, the decisive moment for the autonomous weapons treaty process. A success means the Dialogue produces a clear political signal, backed by a supermajority of participating states, that formal treaty negotiations on lethal autonomous weapons systems must begin at or immediately following the Review Conference, outside the CCW consensus framework if necessary. The 156-state November 2025 UNGA vote demonstrates that the political mass for this exists. The Dialogue's task is to translate that mass into institutional momentum. Second, structural SIDS and Global South inclusion mechanisms. A Dialogue that produces recommendations without embedding participation rights for the states most affected by ungoverned military AI will have reproduced the asymmetry it was convened to address. Success means the Dialogue recommends, at minimum, reserved representation for SIDS regional groupings and dedicated capacity-building financing in any oversight body it proposes. Third, an agreed definition of meaningful human control with operational content. A decade of CCW deliberation has produced the phrase without producing its operationalisation. The Dialogue should adopt a working definition specifying the minimum information, review time, and decision-trail requirements that distinguish genuine human judgment from automation bias, grounded in the cognitive science of decision-making under load. This definition need not be legally binding at Geneva. But it must be specific enough to serve as the baseline for the treaty negotiations that follow. A Dialogue that achieves these three outcomes will have done something the international community has not done in ten years: moved from deliberation to decision.

Social, economic, ethical, cultural, linguistic and technical implications of AI;AI capacity-building;Transparency, accountability, and human oversight;Interoperability of governance approaches;

Our selection is grounded in the empirical record of AI in armed conflict and the structural realities facing Small Island Developing States. Protection and promotion of human rights is the foundational priority because the evidence from Gaza demonstrates that AI targeting systems are already producing civilian casualty rates of 83 percent, with system architectures that make genuine IHL compliance structurally impossible. The right to life, the right to remedy, and the prohibition on arbitrary deprivation of life are not aspirational. They are existing obligations. The Dialogue must treat their violation by AI targeting systems as an immediate governance emergency, not a long-term ethical concern. Transparency, accountability, and human oversight follows directly. The Lavender architecture's twenty-second review cycle, its opaque machine learning algorithms, and its absence of documented decision trails represent a total collapse of accountability. Transparency and oversight are not procedural niceties in this context. They are the legal prerequisites for IHL compliance and the structural conditions for any post-strike accountability process. Social, economic, ethical, cultural, linguistic and technical implications captures the dimension most important to SIDS: the asymmetric distribution of AI's costs and benefits. SIDS face AI-enabled surveillance architectures designed by foreign military-industrial complexes, disinformation campaigns targeting fragile democratic institutions, and displacement of cultural and economic activity, without the institutional capacity to govern or contest these dynamics. This theme must address distributional justice, not only efficiency and innovation. Interoperability of governance approaches is the practical condition for all the above. A world in which China, the EU, the United States, and 156 other states operate incompatible AI governance regimes creates regulatory arbitrage that the most dangerous actors will exploit. Interoperability does not mean uniformity. It means establishing the minimum mutual recognition standards without which no multilateral enforcement mechanism can function.

Three cross-cutting issues of urgent importance fall outside or between the listed themes. The first is AI in armed conflict as a discrete governance domain. None of the eight listed themes directly addresses the use of AI in military targeting, autonomous weapons systems, or AI-enabled warfare. This is the most acute AI governance emergency of 2026, and it is entirely absent from the thematic architecture. The CCW process, the UNGA First Committee, and the Secretary-General's own calls for a binding instrument by end of 2026 all confirm that this is not a peripheral concern. The Dialogue must either create a dedicated thematic track on AI and armed conflict or explicitly integrate it into the human rights and accountability themes with the specificity the issue demands. The second is the necropolitical distribution of AI risk. The thematic framework addresses AI's social and economic implications, but it does not name or address the structural pattern documented across multiple conflict zones: AI systems designed in the Global North are systematically tested, deployed, and exported against populations in the Global South that have no voice in governance deliberations. This is not merely an equity issue. It is a structural feature of the current AI governance architecture that will reproduce itself unless explicitly confronted. The Dialogue needs a mechanism for identifying and addressing this distribution, drawing on both post-colonial analytical frameworks and Rawlsian standards of justifiability to the affected. The third is cognitive warfare and AI-enabled disinformation as a form of hybrid aggression. AI-generated propaganda, deepfakes deployed against democratic institutions, and algorithmic manipulation of public information environments already constitute a form of conflict without armed force. This category has no home in the CCW process, no dedicated treaty instrument, and no mention in the current thematic framework. Its consequences for SIDS, whose democratic institutions are structurally more vulnerable to coordinated external disinformation, are severe and immediate.

Mauritius and the 39 Small Island Developing States we represent sit at the precise intersection where all four of our priority themes converge into a single structural vulnerability. On human rights and accountability, SIDS have no institutional presence in the forums where military AI norms are being set. The CCW GGE in Geneva has 128 High Contracting Parties, but most SIDS lack the permanent diplomatic missions in Geneva required for sustained engagement. Decisions about autonomous weapons, AI-enabled surveillance, and targeting architectures that will define the security environment of the Pacific, Indian Ocean, and Caribbean are being made without a single SIDS voice in the room. When AI-enabled systems developed and tested in conflict zones are exported to regional actors, SIDS bear security consequences they were not consulted on and cannot contest. On social and economic implications, the governance gap is already visible. China's Safe City programme, deploying AI surveillance infrastructure across Africa and the Indian Ocean region, operates without any international governance standard governing data retention, state access, or human rights compliance. Several SIDS host or are proximate to this infrastructure. Their populations are subject to AI systems whose governance architecture is determined entirely by bilateral commercial agreements between their governments and a foreign military-industrial supplier. On transparency and human oversight, UNESCO's 2025 survey found 43 percent of SIDS lack a data governance research framework. Without baseline institutional capacity, SIDS cannot assess, negotiate, or contest the AI systems operating on or around their territory. The opportunity is equally clear. SIDS represent a geographically and culturally diverse coalition of 39 states whose unified voice carries disproportionate moral authority in multilateral settings. The AI Dialogue is precisely the kind of inclusive forum where that authority can translate into structural governance outcomes, if the Dialogue builds the participation mechanisms that make SIDS engagement substantively possible rather than nominally inclusive.

The AI Dialogue occupies an institutional position that no existing forum replicates. It was established by General Assembly resolution outside the CCW consensus veto, without the P5 structure that paralyses Security Council action on military AI, and with an explicit mandate for inclusivity that the GGE process has never operationalised. These features are not incidental. They define a specific and unrepeatable role. That role is to function as the political environment in which the CCW Seventh Review Conference makes its November 2026 decision. The Dialogue does not have binding authority. But it convenes in July, four months before the Review Conference, with 193 member states eligible to participate rather than the CCW's 128 High Contracting Parties. A Dialogue that produces a clear, well-attended, and publicly visible statement of political will for a binding instrument on autonomous weapons will change the political calculus at the Review Conference in ways that no amount of GGE technical deliberation can achieve. Beyond that immediate function, the Dialogue can establish three durable contributions to international cooperation. First, it can create a permanent multi-stakeholder forum that complements rather than duplicates treaty processes, providing a space for civil society, youth, and SIDS to engage substantively between formal negotiating sessions. Second, it can develop and maintain a public registry of national AI governance approaches, creating the baseline transparency from which genuine interoperability can be built. Third, it can commission and publish the first comprehensive assessment of military AI deployments against human rights and IHL standards, using the General Assembly's moral authority to produce the accountability record that no treaty body has yet attempted. The Dialogue's added value is not legislative. It is the construction of the political and epistemic conditions under which binding legislation becomes possible.

The landscape of existing initiatives is rich but fragmented, and fragmentation is itself the core governance failure. The AI Dialogue's primary added value is integration across processes that currently operate in isolation from one another. The CCW GGE on LAWS has produced a decade of technical work, a rolling text, and an emerging consensus on the elements of a binding instrument. The Dialogue should treat this as its foundational technical input, not as a parallel process to be managed separately. Concretely, the Dialogue should invite the GGE Chair to present the rolling text's current state and facilitate a General Assembly-wide discussion of its adequacy, creating political pressure on the Review Conference without duplicating the GGE's technical mandate. The REAIM process, launched by the Netherlands and Republic of Korea in 2023, produced the first multi-stakeholder call to action on responsible military AI, including commitments from over 60 states. The Dialogue should build directly on REAIM's norm development work while extending its reach to the states, particularly in the Global South and SIDS, that REAIM's convening structure did not adequately include. The Global Digital Compact, adopted at the Summit of the Future in 2025, established baseline principles for inclusive digital governance and created the multistakeholder forum the Dialogue can draw on for its civil society and private sector engagement architecture. The UNESCO Recommendation on the Ethics of AI provides the normative foundation for the social and economic implications theme. The Dialogue's specific added value across all of these is threefold. It is the only forum where all 193 UN member states participate as equals. It is the only forum with an explicit youth and civil society inclusion mandate. And it convenes at the precise moment when the gap between these existing frameworks and binding legal obligation must either be closed or acknowledged as the permanent condition of AI governance.