Wikimedia CH

A successful first Global Dialogue would achieve a clear international commitment to AI equity and move from principles to practice. It should outline relevant action areas and provide an actionable roadmap with concrete steps toward AI democratisation while addressing risks and aligning global rules for AI governance. Such a roadmap should recognise knowledge commons as foundational infrastructure for AI democratisation. This should be understood as not just access to AI technologies, but rather as enabling nations and communities to chart their own digital futures. This requires the ability to develop AI systems and solutions, adapt existing technologies to local contexts, and ensure agency in AI development and governance. To this end, the Global Dialogue should reaffirm the Global Digital Compact's commitment to Digital Public Goods (DPGs), including open AI models, open data, and open-source software, and recognise that such goods are produced and maintained within knowledge commons: the communities, platforms, and shared infrastructures whose long-term sustainability is a necessary precondition for a digital future envisioned in the Compact. The Global Dialogue on AI urgently needs to recognize the role of the knowledge commons as critical infrastructure for AI that serves a people-centered digital world and humanity as a whole.

• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Protection and promotion of human rights
• Transparency, accountability, and human oversight
• Open-source software, open data and open AI models

Our perspective draws on Wikimedia CH's work at the intersection of knowledge commons, AI governance, and digital public infrastructure. We propose a thematic agenda centred on knowledge commons as a critical resource and infrastructure for AI. The four priorities address the foundations (open source, data and models), the implications of AI for the digital commons, the governance instruments required, and the normative anchoring of AI governance in international human rights. The priority on open-source software, open data and open AI models recognises that knowledge commons, including collaboratively produced encyclopedic content, open datasets, and cultural heritage, constitute critical infrastructure for contemporary AI, comparable to compute and model architectures. The Global Dialogue needs to address their long-term sustainability and governance, building on commitments to Digital Public Goods and digital commons in the 2020 Roadmap for Digital Cooperation and the Global Digital Compact. A framing centred on knowledge commons as critical infrastructure for AI brings into focus the collaborative development and governance of open resources. The priority on the social, economic, ethical, cultural, linguistic and technical implications of AI acknowledges the systemic transformations AI introduces in the production, distribution and access of knowledge. The focus should be on linguistic, epistemic and cultural plurality, which is a priority for Wikimedia CH in the context of Swiss values and the country's cultural and linguistic diversity. Protecting resilience and diversity is a matter of common concern. The priority on transparency, accountability and robust human oversight reflects the need for governance instruments capable of translating these concerns into obligations. Reciprocity between AI and the knowledge commons, together with transparency, traceability and oversight, is necessary for public trust and responsible AI development. The priority on human rights anchors the agenda in international law: rights of access to information, freedom of expression, and participation in cultural life and scientific progress must be upheld in the development and deployment of AI.

We propose three cross-cutting issues that we consider central to global AI governance: Knowledge commons: Knowledge commons are adjacent to, but distinct from, "open" resources (code, data or models). They include many other types of collaboratively produced, maintained and governed resources. In the context of AI development, sustainability of the knowledge commons is the key issue. Public support is needed, as well as measures that protect the commons from extraction of value, without giving back to the commons. Treating this as a cross-cutting concern allows the Dialogue to consider social, institutional and economic conditions under which open-source software, open data and open AI models are created. Public AI: Public AI extends beyond open-source models to a broader, normative vision of AI grounded in public attributes (openness, interoperability), public functions (advancing public interest goals, including the Sustainable Development Goals), and public ownership or stewardship. Initiatives such as Apertus, the Swiss public multilingual large language model, demonstrate that publicly accountable AI development is feasible and complementary to commercial systems. Public AI and public interest AI development have been previously recognized in the Paris Charter on AI in the public interest. Reciprocity: A norm of reciprocity between developers of AI systems and the communities and infrastructures on which they depend is currently absent from the multilateral debate. Reciprocity cuts across data governance, transparency, sustainability and human rights, and provides a single principle around which obligations toward Digital Public Goods, knowledge commons, and contributing communities can be articulated. Without such a norm, the long-term sustainability of the resources that AI relies upon cannot be secured. Reciprocity is also an important norm when considering global AI divide, and for ensuring democratization of AI capabilities in Global Majority contexts. The Global Dialogue on AI Governance needs to acknowledge the vital value of the knowledge commons as critical infrastructure for AI.

Wikimedia CH speaks from a Swiss perspective shaped by the European context and the global Wikimedia movement. Switzerland's tradition of neutrality, multilateral dialogue, and consensus-building, together with Geneva's positioning as a place of humanity and people-centered technology, make it a natural setting for governance that prioritises collaboration. The most consequential governance gap across these environments is the absence of a coherent framework that recognises knowledge commons as critical infrastructure for AI and provides for their protection, sustainable funding, and democratic governance. Two challenges stand out. First, knowledge commons such as Wikipedia and freely licensed cultural heritage are increasingly extracted as AI training inputs without reciprocity, oversight, or contribution to their long-term sustainability. The Wikimedia Foundation reports that AI crawlers now generate the majority of its highest-cost traffic. Second, public-interest alternatives remain limited, and Public AI is not yet recognised as a distinct policy vision in European or multilateral frameworks. The opportunities lie in aligning multilateral commitments around policies that support not only open-source code and content, but also technical infrastructure, communities, and governance frameworks needed to sustain knowledge commons. Apertus, the Swiss multilingual large language model, shows that public institutions can build AI aligned with these commitments — and reflects Switzerland's capacity to translate its values into infrastructure. Financial mechanisms supporting the commons must also be explored; Wikimedia Enterprise illustrates one workable model of reciprocity with AI companies. The Global Dialogue, hosted in Geneva, should connect national, regional, and multilateral processes around a shared recognition that digital commons are infrastructure warranting public investment and protection.

The AI Dialogue needs to facilitate discussions and collaborations between the different stakeholders working with and affected by AI. Wikimedia CH is an important stakeholder here, representing civil society while also being part of the Wikimedia movement, the Open-Source movement and integral to the knowledge commons. The AI Dialogue needs to ensure that all voices are heard equally, and it needs to call for action against the widening of the global AI divide, but also against the immediate danger of shrinking epistemic diversity if AI does not reciprocate what it takes from the knowledge commons (see Question 12). As transnational corporations become increasingly powerful and as they become more powerful than nation-states, formats such as the AI Dialogue have a special responsibility in amplifying the voices of those with little agency. The way AI is allowed to function now, it consumes the knowledge commons, the information humans have carefully introduced to the digital world, and gives nothing in return. If this human knowledge is not added to the digital anymore, the quality of information online will drastically degrade with harsh consequences for humanity worldwide. In this context, collaboration across all sectors is needed now. Transnational corporations need to be held accountable when it comes to how AI consumes the knowledge commons, but there is also an increasing need for more awareness amongst policy makers globally regarding the prevention of these risks by means of appropriate regulations, and the value of knowledge commons and Digital Public Goods. The AI Dialogue should play a crucial role here. By including the voices of Wikimedia CH and the Wikimedia Movement as a global collaboration platform, the Global Dialogue can ensure that the commitments of UN initiatives such as the Global Digital Compact gain wider recognition.

Initiatives the AI Dialogue should build upon are e.g. the 2020 Roadmap for Digital Cooperation, the Global Digital Compact, and the UNESCO Recommendation on the Ethics of AI, all aimed at a more equitable digital world. The WSIS+20 implementation in 2025 showed the global effort to build more people-centered, inclusive, and development-oriented digital societies, and offered additional guidance on AI. Technology is not neutral and it cannot exist without human interference. Wikimedia projects such as Wikipedia or Wikidata demonstrate how open, participatory projects can sustain knowledge infrastructure at global scale. This infrastructure is increasingly seen as a critical foundation for AI, but it should be more broadly considered as critical infrastructure that serves humanity. The AI Dialogue taking place in Geneva points to certain Swiss values, such as humanity and people-centered tech, transparency and a culture of collaboration. The Dialogue can add value by moving toward coordinated global action, including integrating Digital Public Goods into national AI strategies, promoting interoperability, and supporting investment in shared infrastructure and standards required to support open source communities. The AI Dialogue should connect the mentioned national, regional, and multilateral efforts around a shared understanding that digital commons are essential infrastructures that urgently need public investment and protection. For the AI Dialogue to provide greater coherence, interoperability, and mutual understanding across different stakeholders and processes, the key relation between the knowledge commons and AI must be addressed. The AI Dialogue must also ensure that civil society perspectives are represented. In this context, the voices of the communities that preserve the Wikimedia projects need to be heard.

We suggest making a call for proposals to include civil society- and community-led sessions during the Global Dialogue. There is a critical need for more such perspectives on AI. Wikimedia CH would be pleased to co-chair panels or break-out sessions, e.g. together with other Wikimedia Affiliates, other organizations central to the knowledge commons or likeminded institutions and initiatives such as the Digital Public Goods Alliance. This will help strengthen civil society perspectives and demonstrate the vital role the knowledge commons is playing for the information loop and especially for AI, all whilst not being sufficiently acknowledged and rewarded in return. The impact AI is having on cultures and societies today must urgently get more attention in national and international debates and dialogues. Research perspectives from the intersections of social and cultural sciences as well as tech and AI are crucial here, and Wikimedia CH can make meaningful contributions in this regard.

Communities of the knowledge commons including the Wikimedia communities: Although they ensure the existence of the knowledge commons and therefore of epistemic resilience and diversity in the digital world, their perspectives are not sufficiently included in global discussion about AI governance. The people who contribute to these communities have to be empowered. Information is a fundamental human right, a human right that is threatened by the way AI is allowed to currently function. Stakeholders involved in AI governance need to give voice to communities that defend this right to information. Projects such as Wikipedia or Wikidata, both officially recognized as Digital Public Goods (DPGs) by the UN-endorsed Digital Public Goods Alliance, must be at the center of AI governance discussions. At a time when a few international corporations increasingly dominate the digital world, ensuring that DPGs and the communities behind them are present in governance debates becomes a pressing necessity. Global Majority voices, broadly speaking, also need to be included. The AI divide concerns people of the Global Majority specifically. Wikimedia CH is member of the WikiFranca initiative whose objective it is to connect the "francophonie" of Wikimedia, meaning French-speaking Wikimedia communities. This includes many communities of the African continent and it is an important endeavor for Wikimedia CH to represent these diverse communities and their perspectives on AI in a meaningful way.

We suggest a mix of interactive, multi-stakeholder, and action-oriented formats rather than long plenary-only sessions. The strongest options are ones that match the Dialogue's inclusive, global, and fast-moving structure, especially given the emphasis on broad participation and practical outcomes. Concrete examples are small, mixed-stakeholder roundtables or lighting talks followed by structured Q&A. Action-oriented plans are urgently needed in order to hold the corporations accountable that make AI consume the knowledge commons without reciprocating. The AI Dialogue is the place where such plans must be drafted.

The WSIS+20 implementation in 2025 showed the global effort to build a more people-centered, inclusive, and development-oriented digital society, and offered additional guidance on AI which Wikimedia CH supports. Another key initiative of the WSIS process is the national and international Internet Governance Forums, which have been providing a space for open dialogue on internet governance among civil society, the private sector, the technical community, governments and international organizations. Apertus, the Swiss public multilingual large language model provides one more example. It demonstrates that publicly accountable AI development is feasible and complementary to commercial systems. Another initiative is the DPG Standard of the Digital Public Goods Alliance. They promote AI governance by championing DPGs-open source software, data, and AI systems-as a foundational framework for transparency and equity. The DPG Standard is a primary tool in this effort, which includes specific criteria for AI systems as a category of Digital Public Goods. Practices that offer concrete solutions have also been introduced in the context of the Wikimedia Movement. Wikimedia Enterprise, a service of the Wikimedia Foundation, shows that holding businesses accountable for reusing Wikimedia content at large scale by means of AI is possible. The need for more such initiatives has never been more critical. Wikipedia's community-led governance model itself demonstrates how transparency, human oversight, and participatory decision-making can operate effectively at global scale, while its verifiability and reliable sourcing standards help sustain a high-quality information ecosystem that is critical for AI systems relying on large-scale data. All these initiatives address key problematics of AI and call for a more equitable, sustainable digital society that takes into account linguistic and cultural plurality. The consumption of the knowledge commons by AI is however an issue that remains highly underrepresented.