WE ARE HUMAN

Success would mean three concrete outcomes. First, the Dialogue produces a shared recognition that AI governance is not solely a technical or regulatory matter it is also a cultural and narrative one. The stories AI systems generate, the voices they amplify or erase, and the values embedded in their outputs are governance questions of the first order. A successful Dialogue would explicitly name culture, artistic creation, and civil society creative practice as legitimate domains of governance discussion, not afterthoughts. Second, the Dialogue establishes accountability mechanisms that are practice-based, not only compliance-based. Transparency frameworks for AI systems gain legitimacy when they emerge from the communities using those systems, including artists, educators, and civil society organisations and are embedded in practice from inception rather than imposed retrospectively. A successful Dialogue would surface and validate such models, creating pathways for their wider adoption. Third, the Dialogue advances a concrete agenda for bridging AI divides that goes beyond connectivity and compute. Cultural production who gets to create AI-generated content, in which language, anchored in which heritage is a dimension of AI equity that current frameworks systematically underaddress. A successful Dialogue would commit to measuring and reducing this dimension of the divide. For WE ARE HUMAN, a successful Dialogue would also be one that opens a durable channel between the UN system and civil society organisations working at the operational frontier of ethical AI practice, not only as observers, but as contributors to the ongoing governance architecture.

• Social, economic, ethical, cultural, linguistic and technical implications of AI
• AI capacity-building
• Protection and promotion of human rights
• Transparency, accountability, and human oversight

WE ARE HUMAN is a non-profit association and the convening body of the first international festival dedicated to AI-generated audiovisual works, grounded in an editorial framework aligned article by article with the Universal Declaration of Human Rights. Our four priorities reflect this positioning directly. Cultural and societal implications are our primary lens. AI does not only automate tasks - it reshapes narratives, aesthetics, and collective memory. Governance frameworks must address this dimension explicitly, not absorb it into technical categories. Human rights protection is our editorial architecture. Each of our 30 planned festival editions is anchored to one UDHR article. This structure demonstrates that the UDHR can serve as an operational compass for AI cultural production, not merely an aspirational reference. Transparency and accountability are embedded in our practice through a mandatory instrument we call the Ethics Notebook: a public disclosure document accompanying each work, detailing the AI tools used, biases identified, data provenance, editorial choices made, and carbon footprint generated. This model shows that accountability works best when designed into practice from inception, not layered on as compliance. Capacity-building matters to us because our inaugural edition deploys simultaneously across seven cities on four continents - including São Paulo, Johannesburg, and Taipei with a deliberate centering of the Global South. Bridging AI divides requires attending to cultural production capacity, not only computational infrastructure

Two significant gaps deserve the Dialogue's attention : First: the governance of AI-generated culture as a distinct domain. Current frameworks treat cultural content primarily as a risk category deepfakes, misinformation, copyright infringement. This framing is necessary but insufficient. It ignores the positive governance question: who gets to produce AI-generated culture, in whose language, drawing on whose heritage, under whose editorial authority? The concentration of AI cultural production in a small number of platforms and geographies is a governance problem as serious as the concentration of computational infrastructure and it is not yet named as such in any major international framework. Second: the relationship between AI governance and cultural diplomacy. AI-generated content is increasingly a vector of soft power shaping how nations, peoples, and histories are represented in global information flows. This intersection between AI governance and cultural diplomacy is absent from the current thematic clusters, yet it is operationally active: states and non-state actors are already using AI to project narratives internationally. A future iteration of the Dialogue should address the governance implications of AI as a diplomatic and geopolitical instrument, with particular attention to the perspectives of countries and communities whose cultural heritage is most at risk of misrepresentation or extraction. WE ARE HUMAN's work directly addresses both gaps through its UDHR-anchored editorial architecture and its multi-continental deployment - and we would welcome the opportunity to contribute further to the Dialogue's reflection on these emerging dimensions.

The civil society and cultural sector faces a governance gap that is structural, not incidental. AI tools capable of generating images, voices, narratives, and audiovisual works at scale are now accessible to any organisation with minimal resources. Yet the frameworks governing their use in cultural production, in terms of transparency, rights, provenance, and accountability ,remain almost entirely absent at the international level. For organisations working at the intersection of AI, human rights, and cultural diplomacy, this gap creates both acute challenges and genuine opportunities. The central challenge is one of legitimacy. When civil society organisations use AI in cultural production — even deliberately, critically, and transparently — they operate without recognised standards or shared vocabulary. There is no international framework that validates ethical AI cultural practice, distinguishes it from extractive or manipulative uses, or provides the governance language needed to build trust with institutional partners, funders, and audiences. This absence weakens the credibility of responsible actors while offering no barrier to irresponsible ones. A second challenge is geopolitical asymmetry. The cultural narratives produced and distributed through AI systems largely originate in a small number of technologically advanced countries. Communities in the Global South — whose oral traditions, heritage, and living cultures are among the richest in the world have limited capacity to shape how AI represents them, and limited recourse when it does so inaccurately or exploitatively. Governance frameworks that do not address this asymmetry will entrench it. The opportunity, however, is real. Civil society organisations already developing practice-based accountability models such as embedded transparency instruments, multi-continental diffusion architectures, and editorial frameworks anchored in human rights offer the Dialogue concrete, tested approaches that complement top-down regulatory frameworks with bottom-up governance practice.

The AI Dialogue's most distinctive potential contribution to international cooperation is one that no bilateral agreement, regional framework, or industry standard can replicate: it is the only universal forum where governments and civil society sit at the same table, with equal entitlement to shape the agenda. That universality is not merely procedural it is substantive. AI governance challenges that are genuinely global in their consequences, but unequal in their distribution, require a forum where the countries and communities most affected by those consequences have standing. The Dialogue can be that forum, but only if it resists the tendency to treat governance as primarily a technical coordination problem among advanced economies. For the cultural and civil society sector, the Dialogue's role in international cooperation is threefold. First, it can establish shared vocabulary. One of the most significant obstacles to international cooperation on ethical AI in cultural production is the absence of common definitions for transparency, for accountability, for meaningful human oversight. The Dialogue can surface and validate terminology that travels across jurisdictions and sectors. Second, it can create recognition pathways. Civil society organisations and cultural institutions working responsibly with AI lack any international mechanism to have that practice recognised, validated, or connected to others doing equivalent work in different contexts. The Dialogue can build those bridges. Third, it can anchor norm formation in practice rather than aspiration. The most durable international norms emerge from documented practice, not declaration. By centering the experiences of practitioners including artists, educators, and civil society organisations alongside governments and industry, the Dialogue can generate governance insights grounded in what actually works.

Several existing frameworks and initiatives offer the AI Dialogue genuine foundations to build upon rather than duplicate. UNESCO's 2021 Recommendation on the Ethics of AI is the most directly relevant existing instrument for the cultural and civil society sector. It calls explicitly on member states to harness cultural and creative expression as vectors of ethical reflection on AI, and it provides a normative basis for the practice-based accountability models that civil society organisations are already developing. The Dialogue should treat the UNESCO Recommendation not as a reference document but as an operational framework to be advanced and monitored. The Global Digital Compact, which mandated the Dialogue itself, established commitments on digital inclusion and the responsible use of emerging technologies that remain largely unimplemented in the cultural sector. The Dialogue should take stock of that implementation gap and identify civil society actors already working to close it. At the practitioner level, initiatives that embed transparency and human rights accountability directly into AI creative processes — such as the Ethics Notebook framework developed within the WE ARE HUMAN festival — represent the kind of bottom-up governance practice that international frameworks rarely acknowledge but consistently need. Connecting these practices to the Dialogue's intergovernmental process would strengthen both. Finally, the AI for Good Global Summit, in whose margins the first Dialogue session will be held, provides an immediate connection to the technical and innovation communities. The Dialogue's added value relative to that Summit is precisely its human rights and civil society dimension that differentiation should be preserved and strengthened, not blurred. The Dialogue's unique contribution is not to replace existing mechanisms but to provide the universal, multi-stakeholder home they currently lack.

The AI Dialogue will only generate durable governance outcomes if it treats civil society not as an observer category but as a co-producer of knowledge. Governments bring regulatory authority. Industry brings technical capacity. Civil society brings something neither can replicate: direct accountability to the communities most affected by AI's consequences, and documented practice in navigating those consequences without institutional protection. In practical terms, this means the Dialogue's format should reserve dedicated speaking time for civil society practitioners in each thematic session, not only in open floor segments where institutional voices naturally dominate. It also means that written inputs from civil society organisations should be synthesised and reflected in the Co-Chairs' summary, not treated as background documentation. The two-day Geneva format is necessarily compressed. To make it generative rather than ceremonial, thematic breakout sessions should be structured around concrete case studies and documented practices, with practitioners presenting alongside government delegates. The goal of each session should be to surface at least one replicable model or actionable recommendation, not only to map the landscape of concerns. Between sessions, the Dialogue should maintain a living repository of civil society contributions, openly accessible, which delegates and stakeholders can consult and build upon ahead of the 2027 New York session.

The most consequential underrepresentation in global AI governance discussions is not demographic it is epistemological. The communities whose cultural heritage, oral traditions, and living knowledge systems are most exposed to AI's extractive logics are systematically absent from the rooms where governance norms are set. Artists, storytellers, and cultural practitioners from the Global South hold deep expertise in exactly the questions AI governance must answer: whose knowledge counts, whose voice is authentic, whose heritage can be reproduced and by whom. That expertise is not recognised as governance knowledge in current frameworks. Young people are similarly absent. They are the primary users of AI-generated content, the generation that will live longest with the governance decisions made now, and in many of the countries represented in Geneva, they constitute the majority of the population. Their perspectives should be structurally integrated, not invited as a gesture. Finally, artists and cultural organisations working with AI in good faith occupy a distinctive position: they use these tools critically, transparently, and accountably, and they face the governance gaps directly in their daily practice. The WE ARE HUMAN festival is built on the conviction that this community has concrete contributions to make to governance, not only to aesthetics. The Dialogue should create pathways for that contribution.

The most effective formats for meaningful engagement are those that make governance stakes visible and tangible, not only discussable. Opening each thematic session with a short work — a film, a testimony, a documented creative process — that concretely illustrates the governance challenge under discussion would shift the register from abstract policy to lived consequence. This is not decoration. It is a method for grounding intergovernmental exchange in the realities it is meant to address. Structured peer exchange between practitioners from different regions, facilitated in small groups alongside the plenary sessions, would allow the kind of direct knowledge transfer that large-format meetings cannot produce. A civil society organisation running an AI ethics programme in Johannesburg and one doing equivalent work in Taipei have more to learn from each other than from another panel of experts — the Dialogue should create the conditions for that exchange.

The most instructive examples of effective AI governance are not found in regulatory texts - they are found in organisations that have made accountability operational, at the point of creation, with real works and real audiences. WE ARE HUMAN is developing one such example directly. Our festival commissions AI-generated audiovisual works anchored to the Universal Declaration of Human Rights, and requires every work presented to be accompanied by an Ethics Notebook: a public document disclosing the AI tools used, the biases encountered and the choices made to navigate them, the cultural provenance of the data drawn upon, and the carbon cost of the generation process. This instrument has been tested in production with works addressing African oral heritage and post-colonial identity, created by artists from South Africa, for international presentation. It functions. It is replicable. It costs nothing to adopt beyond the editorial commitment to do so. What this model demonstrates concretely is that human oversight does not require an external auditor or a regulatory body to be meaningful. It requires that the human beings making creative decisions document those decisions, expose them to public scrutiny, and treat transparency as a condition of legitimacy rather than a burden of compliance. A second concrete practice worth the Dialogue's attention is multi-continental simultaneous diffusion as an equity instrument. Rather than exporting AI cultural content from technologically advanced countries to others, our festival architecture places cities in the Global South - Johannesburg, São Paulo, Taipei - as co-equal programme sites from the first edition. The governance insight here is simple: inclusion must be structural, not aspirational. It must be designed into the architecture of the initiative before the first work is shown, not added later as outreach. These are not proposals. They are practices already underway, available to inform the Dialogue's recommendations. Contact: Michael Swierczynski, President - mski@wearehuman.foundation / michaelswi06@gmail.com