Generative AI Commons, LF AI & Data

I am an open-source strategist and community builder. I have served on several open-source foundation boards and helped build global communities, including the OpenInfra Foundation and projects within the Linux Foundation's CNCF. I currently serve as an elected board director at LF AI & Data and co-chair the Generative AI Commons. Across these communities, we bring together technologists worldwide who are committed to building open, transparent, and responsible AI that serves humanity. In this Global Dialogue on AI Governance, I would like to emphasize the importance of actively incorporating the open-source community into our efforts. Open source is not only a development model—it is a governance model. It enables transparency by design, supports inclusivity by lowering barriers to participation, and accelerates innovation through shared infrastructure. Open-source ecosystems already provide practical mechanisms aligned with global AI governance goals, including community-driven standards, open evaluation benchmarks, shared safety tools, and reproducible research. These contribute directly to accountability, interoperability, and trust. Moreover, open-source communities offer neutral, multi-stakeholder spaces where governments, industry, academia, and civil society can collaborate in practice. This is particularly valuable for the United Nations in building governance frameworks that are globally representative and technically grounded. By engaging more directly with open-source initiatives, the UN can help scale trusted digital public goods, strengthen global capacity, and ensure governance is informed by real-world implementation. I encourage this dialogue to consider structured pathways to involve and support the open-source community as key partners in shaping global AI governance solutions.

• Open-source software, open data and open AI models
• Transparency, accountability, and human oversight
• Safe, secure and trustworthy AI
• Interoperability of governance approaches

Open source software, open data, and open models are foundational to effective and inclusive global AI governance because they embed transparency, accessibility, and collaboration into the technology itself. First, they enable transparency and accountability. Open systems can be inspected, audited, and tested by independent experts across the world, which is essential for building trust in AI systems that impact societies at scale. This reduces reliance on opaque, proprietary systems and helps identify risks, biases, and safety issues early. Second, they promote equity and inclusion. Open access lowers barriers for participation, allowing researchers, developers, and institutions-especially from the Global South-to contribute to and benefit from AI development. This helps ensure that AI governance is not shaped solely by a few dominant actors, but reflects diverse perspectives and needs. Third, they accelerate innovation and interoperability. Shared tools, datasets, and models prevent duplication, enable faster iteration, and support the development of common standards. This is critical for addressing global challenges that require coordinated, cross-border solutions. For the United Nations, including the open-source community in global AI governance is both practical and strategic. These communities already function as distributed, multi-stakeholder ecosystems that bring together industry, academia, civil society, and governments in collaborative problem-solving environments. They also produce digital public goods that can be adopted and adapted globally. By engaging the open-source community, the UN can ground its governance efforts in real-world technical expertise, foster global capacity-building, and ensure that AI systems remain transparent, interoperable, and aligned with public interest values.

global open collaboration in AI innovation, knowledge sharing, and solution building

I see the top AI models are proprietary and closed sourced. This is very dangerous for the following reasons: First, lack of transparency is a major concern. When the most powerful AI systems are closed, their training data, design choices, and limitations are not fully visible. This makes independent auditing difficult and limits the ability of governments, researchers, and civil society to assess bias, safety, or unintended consequences. Second, concentration of power becomes an issue. A small number of companies control the most advanced models, which can create imbalances in economic influence, information access, and technological sovereignty. This can leave many countries—especially in the Global South—dependent on external providers for critical infrastructure. Third, there is limited accountability. If systems are opaque, it becomes harder to assign responsibility when harms occur, whether related to misinformation, discrimination, or system failures. External stakeholders often have little recourse beyond trusting the provider. Fourth, restricted access and participation can slow broader innovation and capacity-building. Researchers, startups, and public institutions may face barriers to experimenting, adapting, or improving these systems for local needs. Finally, systemic risk increases when widely used technologies cannot be independently verified or adapted. If vulnerabilities or failures emerge, they may affect many sectors simultaneously without a transparent path to mitigation. For these reasons, a more balanced ecosystem—where open and transparent approaches complement proprietary systems—can help ensure that AI development remains accountable, inclusive, and aligned with the public interest.

The AI Dialogue can play a catalytic role in advancing international cooperation on AI governance by positioning open source as a shared, practical foundation for collaboration. First, it can recognize open source as digital public infrastructure. By elevating open-source software, data, and models as global public goods, the Dialogue can encourage coordinated investment and adoption across countries, especially to support developing economies. Second, it can convene multi-stakeholder collaboration. Open-source communities already bring together governments, industry, academia, and civil society. The Dialogue can serve as a bridge, formally integrating these communities into policy discussions and ensuring governance is informed by real technical practice. Third, it can promote common standards and interoperability. Through open collaboration, the Dialogue can support the development of shared benchmarks, safety tools, and governance frameworks that are transparent and widely adoptable across jurisdictions. Fourth, it can support capacity-building and knowledge sharing. By leveraging open-source ecosystems, the Dialogue can help countries build local AI capabilities, reduce dependency on a few providers, and enable more equitable participation in the global AI landscape. Finally, it can pilot collaborative governance models. The Dialogue can encourage joint initiatives—such as open evaluation platforms, shared datasets, or safety toolkits—that demonstrate how open, cooperative approaches can address global AI risks. In this way, the AI Dialogue can move beyond principles and help operationalize international cooperation through open, inclusive, and actionable mechanisms grounded in the open-source ecosystem.

Within LF AI & Data, we have established the Generative AI Commons - https://genaicommons.org/, an open community advancing practical approaches to AI governance. One of our key workstreams focuses on Responsible Generative AI with a high value output - Responsible AI Framework (RGAF), where we address nine core tenets that define trustworthy and ethical AI systems. This work is not only conceptual—we are actively identifying and supporting open-source projects that can implement, test, and verify these principles in practice. In addition, we have developed a Model Openness Framework (MOF), which provides a structured way to assess and compare the degree of openness of AI models across dimensions such as data, code, weights, and documentation. This helps bring much-needed clarity and consistency to discussions around transparency and accountability. These efforts demonstrate how open-source communities can translate governance principles into operational tools, benchmarks, and implementations. They also create shared resources that can be adopted globally, reducing fragmentation and accelerating alignment. There is strong potential for collaboration with the United Nations. Initiatives like these can support global AI governance by offering practical frameworks, open standards, and verifiable tooling that are developed through inclusive, multi-stakeholder processes. By engaging with and building on such open efforts, the UN can help ensure that AI governance is not only principled, but also actionable, measurable, and globally accessible.

It is important to continue building on forums like today's dialogue by creating more regular and inclusive opportunities for engagement. Increasing the frequency of open, virtual consultations can significantly broaden global participation, particularly from diverse technical communities that may not always be present in traditional policy settings. In addition, I would encourage the creation of a dedicated open-source track at the upcoming AI for Good Summit in Geneva. This would provide a practical platform to showcase real-world implementations, foster collaboration, and connect policy discussions with working technologies. At present, much of the engagement within ITU has focused on the standards community. While this is essential, it is equally important to actively involve the open-source ecosystem, which plays a critical role in building and deploying the technologies that standards aim to guide. Organizations and communities such as LF AI & Data and the Generative AI Commons represent global, multi-stakeholder networks with deep technical expertise and a strong commitment to open, responsible AI. By more intentionally engaging the open-source community, ITU can strengthen the link between standards, implementation, and innovation ensuring that global AI governance is not only well-defined, but also widely adopted and operational in practice.

Open Source Community which are actively working on open source projects such as Linux Foundation, Eclipse Foundation, Apache Foundation, etc. don't have presence at UN's AI initiatives while they are driving some of the most widely used AI technologies and solutions

Online forum is a great start. Regular publication of AI discussions via the AI Dialogue platform. Open Source and Open Science projects that address global AI challenges and opportunities

Some of our work created by the Generative AI Commons open source community - Model Openness Framework - https://isitopen.ai/ - Responsible Gen AI Framework - https://lfaidata.foundation/blog/2025/03/19/responsible-generative-ai-framework-rgaf-version-0-9-now-available/