WTL Governance

A successful first Dialogue would produce three concrete outcomes. First, a Co-Chairs' Summary that does more than synthesize stakeholder views: it should identify the two or three structural AI governance risks where coordinated international attention is most urgent, name them specifically rather than gesturing at categories, and assign each to specific UN bodies for follow-through. Vague summaries become talking points, whereas specific summaries become workstreams. Second, the Dialogue should produce credible continuity mechanisms between the 2026 and 2027 sessions. The pace of AI capability development and labor market change exceeds what an annual cadence can capture. Lighter thematic follow-up tracks, particularly on labor and human capital, on interoperability, and on capacity-building, would allow the Dialogue to translate the Scientific Panel's evolving evidence base into governance discussion as it develops, rather than waiting twelve months for the next plenary. Third, the Dialogue should demonstrate genuine multi-stakeholder character without reducing to a procedural checkbox. Civil society, academia, the private sector, and the technical community each see different parts of the picture; an effective Dialogue will surface where their views diverge from intergovernmental positions, not just where they converge. The Scientific Panel's first report will be most useful if the Dialogue debates it openly rather than treating its findings as settled. Beyond these three outcomes, the inaugural Dialogue carries an additional burden: establishing legitimacy in the face of the geopolitical fracturing visible in the February 2026 Scientific Panel vote and the ongoing absence of US engagement in multilateral AI governance. A Dialogue perceived as serious, evidence-driven, and operationally consequential will be harder to sideline.

• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Interoperability of governance approaches
• Protection and promotion of human rights
• Transparency, accountability, and human oversight

The four selected reflect WTL's analytical and operational positioning across the AI governance landscape. The implications cluster houses what WTL views as the most urgent and under-recognized structural challenge currently visible across jurisdictions: the disruption of the entry-level career ladder in AI-exposed sectors. Marginal labor data from the United States, France, the United Kingdom, and South Korea show convergent patterns of declining youth employment in AI-exposed cognitive work, even as aggregate indicators remain healthy. Frameworks built only on aggregate data will miss this dynamic entirely. Interoperability reflects the practical reality that AI systems and developers now operate across jurisdictions while enforcement remains national. WTL's regulatory tracking across the United States, the European Union, and the Asia-Pacific has documented growing divergence between binding regimes (the EU AI Act, China's measures), voluntary frameworks (Japan, the Hiroshima AI Process), and sector-specific regulation (the United States). Interoperability is the mechanism by which divergent approaches can coexist without imposing compliance costs that exclude the small enterprise from participation. Capacity-building addresses the talent pipeline dimension of the entry-level squeeze. As AI absorbs the entry-level tasks through which human expertise has historically been built, the developing world risks being locked out of the senior expertise pipeline entirely. Capacity-building must include not only access to AI tools and infrastructure, but also the developmental architecture that produces governance, technical, and policy professionals over time. Transparency, accountability, and human oversight is foundational. Recent research has formally established that AI hallucination cannot be eliminated; human oversight is not a transitional measure but a permanent structural requirement. Accountability frameworks must reflect that reality, not aspire to a fully autonomous future that the underlying technology cannot deliver.

Three cross-cutting issues sit awkwardly within the proposed thematic clusters and warrant explicit attention. First, AI infrastructure at the intersection of compute, energy, water, and critical minerals. The Secretary-General has flagged the geopolitical and environmental tensions created by AI's rising resource demand; UNCTAD's 2024 reporting documented that AI data centers consumed as much electricity as France in 2022 and were projected to double by 2026. This dimension sits across "implications" (economic and environmental), "capacity-building" (Global South access depends on infrastructure), and trustworthiness (resilience of supply chains). Treating it as the residual concern of any single cluster will mean it receives the residual attention of all of them. Second, the apprenticeship and skill-transfer architecture by which human expertise is produced. The disruption of the entry-level career ladder, documented in marginal labor data across the United States, France, the United Kingdom, and South Korea, is not solely a labor question. It is a question of how societies maintain the institutional knowledge required for governance, technical work, and professional practice. The expert-novice apprenticeship model has been the primary mechanism for human skill transfer for centuries; AI is now disrupting it without an evident replacement. This is upstream of capacity-building and downstream of trustworthiness, but is not adequately captured by either. Third, the information ecosystem. Synthetic media, AI-generated content, and the erosion of public information integrity sit between "trustworthy" AI and "human rights" but are not adequately captured by either alone. The UN Global Principles for Information Integrity (June 2024) established a foundation; the Dialogue should consider whether the cluster structure leaves it sufficient room. Information integrity is, in many democracies, the most politically salient AI governance challenge currently, and stakeholder submissions from civil society are likely to surface it repeatedly.

WTL Governance operates across the United States, the European Union, and the Asia-Pacific, with the firm based in Tokyo and Wyoming. Three observations from that vantage point. On the implications cluster, the workforce evidence is now arriving faster than governance frameworks can absorb it. As mentioned in previous answers, marginal labor data from the United States, France, the United Kingdom, and South Korea show convergent declines in youth employment in AI-exposed cognitive work, even as aggregate indicators remain healthy. The challenge is that domestic labor ministries, education systems, and AI regulators rarely coordinate; the opportunity is that the Dialogue can surface this as a structural risk warranting joint attention from the ILO, UNESCO, and the Scientific Panel before national responses fragment further. On interoperability, organizations operating across jurisdictions face binding regimes (the EU AI Act), voluntary frameworks (the Hiroshima AI Process and Japan's guidance-based approach), and sector-specific oversight (the United States), with growing divergence between them. The challenge is that smaller economies and Global South participants cannot bear the compliance cost of three or more parallel regimes; the opportunity is that the Dialogue is uniquely positioned to identify which substantive elements admit harmonization and which reflect genuine sovereign choice that should be respected rather than overridden. On capacity-building and on transparency and oversight, the same structural pattern recurs across regions. As AI absorbs the entry-level tasks through which expertise has historically been built, the pipeline that produces governance, technical, and policy professionals erodes upstream of every other workstream. Recent research has formally established that AI hallucination cannot be eliminated, making human oversight a permanent structural requirement that depends on the very expertise the entry-level squeeze is hollowing out. The opportunity is to address these as a connected system rather than as parallel concerns.

The Dialogue's distinctive value is structural: it is the only forum in which all 193 UN Member States have a seat at the AI governance table on equal footing. Every other consequential venue, including the G7's Hiroshima AI Process, the AI Safety Summit series, the OECD-hosted GPAI, and the AI Safety Institute network, operates through more selective membership. Each has produced substantive output; none speaks for the world's population. The Dialogue can play three roles that no other forum is positioned to play. First, it can function as the universal forum where the substantive work of more selective venues is presented, debated, and connected. The HLAB-AI documented in 2024 that 118 countries, overwhelmingly in the Global South, were entirely excluded from every major AI governance initiative. The Dialogue is the mechanism by which their governments and stakeholders can engage with the frameworks being built without them, and by which those frameworks can be tested against perspectives the original drafters did not include. Second, the Dialogue can serve as the institutional bridge between the Independent International Scientific Panel's evidentiary mandate and the policy choices that flow from it. Scientific assessments without political deliberation become technical reports; political deliberation without scientific assessment becomes posturing. The Dialogue's pairing with the Panel, both established by Resolution 79/325, gives it a structural advantage no other forum has. Third, the Dialogue can identify where international cooperation is genuinely needed and where it is not. Not every AI governance question requires multilateral coordination; some are properly resolved domestically or regionally. A Dialogue that distinguishes the cooperation-essential from the cooperation-optional will be more useful than one that treats all AI governance issues as global by default.

The AI governance landscape is institutionally dense. Accordingly, the Dialogue should build on rather than duplicate existing infrastructure. On normative frameworks, the UNESCO Recommendation on the Ethics of AI (2021), adopted unanimously by 193 Member States, remains the foundational instrument and the explicit reference point of the General Assembly's first AI resolution. The OECD AI Principles (2019, updated 2024) anchor every major G7 AI instrument. The Council of Europe Framework Convention on AI (2024) is the only legally binding international AI instrument and includes the United States as a signatory. The Dialogue's added value is to situate these instruments within a single coherent landscape that all Member States can engage with, rather than maintaining the current architecture in which different governments engage with different frameworks. On institutional partnerships, the Dialogue should connect formally with the OECD-hosted GPAI for applied policy research, the ILO for labor and workforce evidence, UNESCO for ethics and education, the WHO for healthcare AI, and the AI Safety Institute network for technical assessment capacity. Each has substantive depth the Dialogue cannot replicate; the Dialogue's added value is convening across them, not duplicating them. On parallel processes, the Dialogue should engage with the Hiroshima AI Process and AI Safety Summit series as substantive inputs, recognizing that frameworks built outside the UN system contain operational lessons the Dialogue can absorb. The added value here is integration: bringing the technical and operational work of selective venues into deliberation with the broader Member State community. The Dialogue's distinctive contribution lies in three places no existing mechanism occupies: universal Member State participation, structural pairing with the Scientific Panel's evidentiary mandate, and the political legitimacy that flows from General Assembly establishment under Resolution 79/325. The Dialogue should lean into these advantages rather than competing with venues that have other strengths.

Different stakeholder categories contribute most effectively when the Dialogue's structure reflects what each can uniquely offer. Member States contribute political legitimacy and the binding-instrument capacity that voluntary frameworks lack. The high-level governmental segment is the appropriate venue for this contribution. Statements should be substantive rather than ceremonial; the Co-Chairs may wish to encourage delegations to address specific cooperation questions rather than delivering general position statements that summarize national approaches. Civil society contributes the perspective of populations affected by AI deployment, particularly those underrepresented in the development of the technology itself. Civil society participation should not be confined to formal interventions; the most useful contributions often emerge from cross-stakeholder exchange where civil society can challenge industry and government positions directly. The technical community contributes operational knowledge of how AI systems actually behave, which often diverges from how policymakers and the public perceive them. The Scientific Panel addresses this at the highest level, but the Dialogue should also include working-level technical voices in thematic discussions, particularly on safety, transparency, and oversight. Industry contributes implementation experience and resource capacity, but its participation requires careful structuring to avoid disproportionate influence. Industry voices should be included alongside, not in place of, civil society and academic perspectives on the same questions. Academia contributes independent analysis and longer time horizons than other stakeholders typically offer. Similar to government, academic participation should be substantive, not decorative. On structure: the proposed two-day format gives high-level framing room to land, but thematic discussions are where substantive cooperation insights will emerge. Co-chairing each thematic session with a Member State and a relevant stakeholder is the right design. The Co-Chairs may wish to ensure that the thematic discussion summaries reach the Co-Chairs' Summary directly, rather than being filtered through the high-level segment's framing.

Three categories of voice are systematically underrepresented in current AI governance discussions. First, Global South governments and stakeholders. As mentioned, the HLAB-AI documented in 2024 that 118 UN Member States, the overwhelming majority in the Global South, were entirely excluded from every major AI governance initiative the Body surveyed. The Dialogue is structurally well-positioned to address this, but inclusion requires more than open invitations. It requires investment in participation infrastructure: travel support, language access in all six UN languages, time-zone-conscious scheduling for virtual components, and Co-Chairs' Summary drafting that meaningfully reflects perspectives from outside the dominant Anglophone discourse. Second, workers and worker representatives. AI's labor impacts are emerging across the global economy, with marginal labor data showing convergent patterns of declining youth employment in AI-exposed sectors across the United States, France, the United Kingdom, and South Korea. Yet workers themselves are largely absent from AI governance forums. The ILO's Constitution requires tripartite participation by governments, employers, and workers; the Dialogue should consider engaging with this tripartite structure rather than treating labor as a single stakeholder category alongside others. Third, communities affected by AI deployment in high-stakes domains: healthcare, criminal justice, social welfare, immigration, and education. Their experience differs from that of developers, regulators, and even the civil society organizations that often speak on their behalf. Inclusion requires deliberate engagement with affected-community organizations and lived-experience expertise, not simply broader invitations to existing civil society networks. Practical mechanisms for inclusion include: dedicated rapporteur roles to surface underrepresented perspectives in thematic sessions; structured input from regional intergovernmental organizations to channel views from countries unable to send delegations; and explicit acknowledgment in the Co-Chairs' Summary of which perspectives were heard and which remain underrepresented despite efforts to include them.

Four format choices would meaningfully improve the Dialogue's substantive output. First, structured pre-Dialogue evidence rounds. Stakeholders submit specific evidence on each thematic cluster ahead of the session, the Co-Chairs publish a synthesis, and the in-person discussions begin from that shared evidentiary baseline rather than competing assertions. This is how the IPCC handles climate evidence and how the OECD handles peer review; it produces substantively richer discussion than venues that begin from zero each time. Second, cross-stakeholder breakout discussions on specific governance challenges, structured to require interaction across stakeholder categories rather than parallel monologues within them. A breakout on workforce impacts, for example, should bring together a labor ministry, an industry voice, a worker representative, and an academic on the same topic, with a moderator empowered to surface disagreement rather than smooth it over. Third, formal recognition that the Dialogue's most valuable substantive work may happen at thematic working level, not at the plenary. The high-level segments serve political legitimation; the thematic discussions surface the operational insights that should flow into the Co-Chairs' Summary. Resourcing the thematic sessions with adequate time, capable rapporteurs, and direct authorship channels into the Summary will produce a more substantive output. Fourth, continuity infrastructure between the 2026 and 2027 sessions. Lighter thematic follow-up tracks, organized around specific governance questions and coordinated with existing UN bodies, would allow stakeholder engagement to deepen rather than reset. The pace of AI capability and labor market change exceeds what an annual cadence can capture. The underlying principle: the Dialogue should be designed for substantive output, not procedural completeness. Formats that prioritize broad participation over substantive depth produce summaries that no one acts upon.

Several existing instruments and approaches offer concrete templates the Dialogue can build upon. On binding regulation, the EU AI Act's risk-tiered framework provides a working example of proportionate regulation: prohibited practices, high-risk systems with conformity assessments, limited-risk systems with transparency obligations, and minimal-risk systems left to general law. The Council of Europe Framework Convention on AI (2024), the only legally binding international AI instrument, demonstrates that cross-jurisdictional binding cooperation is achievable; the United States is a signatory. On sectoral governance, the WHO's January 2024 guidance on large multi-modal models in healthcare provides over 40 specific recommendations covering diagnosis, clinical care, administrative tasks, medical education, and drug development. It demonstrates that operationally detailed guidance is achievable in high-stakes domains. The IMO MASS Code for autonomous vessels offers a phased compliance roadmap (non-mandatory adoption, experience-building phase, mandatory transition by 2032) that other sectors could adapt. On voluntary frameworks with practical traction, the Hiroshima AI Process Reporting Framework, administered by the OECD, asks organizations developing advanced AI systems to demonstrate Code of Conduct implementation through structured questionnaires. Reporting Framework v2.0 is now in pilot. The mechanism translates principles into observable practice without requiring binding enforcement. On capacity-building, UNESCO's Readiness Assessment Methodology has been implemented in over 70 countries with more than 17,000 stakeholders consulted. It provides governments with diagnostic tools to assess national AI readiness and identify priority interventions. On technical cooperation, the International Network of AI Safety Institutes, launched in November 2024 with members from ten jurisdictions, conducts joint testing exercises and aligned risk assessments. It demonstrates that multilateral technical cooperation on AI safety is possible across geopolitically diverse participants. Each offers operational lessons. The Dialogue's task is connection, not duplication.