March 5, 2026
Ensuring “Openness” in AI is a Beacon of Trust, Not a Slogan: Reflections from the India AI Impact Summit 2026
Author: Lea Gimpel, Director of Policy & AI Lead and Amreen Taneja, Standards Lead
Returning from New Delhi after the India AI Impact Summit 2026, it is clear that the global AI conversation has reached a turning point. As the first major AI summit held in the Global South, the event centred on impact, democratisation, and the specific needs of developing economies. For the Digital Public Goods Alliance, the week provided a vital platform to discuss the foundational role of digital public goods in AI development and to connect with like-minded partners committed to supporting the open AI ecosystem for a more equitable digital future.
Openness: A Foundation, Not a Finish Line
At a time when data from HuggingFace suggests that we are witnessing a collapse in transparency in AI development, openness emerged as a recurring theme at the AI Impact Summit and discussed as a fundamental building block of AI infrastructure for countries, the private sector and end users.
At the DPGA, we believe openness is the necessary starting point for achieving transparency, trust, sovereignty, and the democratisation of AI resources. However, without integrated best practices, guardrails, and strong governance, "openness" remains a hollow term, leaving it open to use in ways that may not always advance the public good. Since its inception, the DPG Standard, as a community-driven benchmark, has addressed these aspects through specific indicators on "do no harm by design" and "adherence to privacy and applicable laws".
The DPG Standard: Verifying Public Benefit
The DPG Standard is grounded in the UN Secretary-General’s definition: digital public goods are open-source software, open datasets, open AI systems, and open content collections that adhere to privacy laws, follow best practices, and help attain the Sustainable Development Goals (SDGs). With this, the DPG Standard moves the definition from “openness” as a standalone characteristic to “openness” as a feature complemented by other features, with a specific objective: making the world a better and more equitable place.
The DPG Standard allows us to move beyond the "open source" label by providing a verifiable framework that ensures technology is not only technically accessible but also safe, inclusive, and designed for public benefit. As a meta-standard that also encourages adherence to other standards and best practices, as per Indicator 8, the DPG Standard ensures that AI safety best practices are incorporated through continuous consultation with the community and experts.
The Threefold Intersection of DPGs and AI
The intersection of DPGs and AI is threefold:
- AI systems as a category of digital public goods.
- DPGs' use of AI to enhance and run their products.
- DPGs are trusted resources that can improve responsible AI development —such as open data, tooling, and software (DPG4AI).
This third category includes resources such as compar:IA (a chatbot comparison arena developed by the French government to reduce cultural and linguistic biases in LLMs); simpleAudit (framework for AI safety testing); and the Mozilla Common Voice dataset (multi-lingual dataset for language technology).
DPG4AI was also the topic of the DPGA Secretariat’s mainstage session, “Digital Public Goods for Global AI Equity,” at the Summit in India which included John Dickerson, CEO of mozilla.ai, Ayah Bdeir, CEO of CurrentAI, Chenai Chair, Director of the Masakhane African Languages Hub and Urvashi Aneja, Founder and Executive Director of Digital Futures Lab. There the discussion focused on persisting challenges such as the shortage of high-quality open data, market concentration, and the skills gap to develop and deploy AI in ways that benefit public interest. Panellists explored the role of openness in supporting the digital sovereignty of nations and people and raised questions about the fine balance between championing openness while avoiding corporate capture and ensuring that communities benefit from their own data collection work. Having CurrentAI, one of the main deliverables of last year’s French AI Action Summit, on the panel with Ayah Bdeir also enabled us to connect the dots between Paris and New Delhi, underscoring the essential role of open, global collaboration in advancing AI that serves the many, not the few.
Trust and Safety in Local Open-source AI solutions
Representing the DPGA Secretariat, we also had the privilege of moderating a session titled "Harnessing Open Source AI for Inclusive Economic Development," hosted collaboratively by GIZ FAIR Forward, NASSCOM, and Digital Futures Lab. During this session, Yasha Khandelwal, CEO & Founder of Tech4Biz, shared that open-source initiatives are effectively reducing access barriers for local developers. However, the Indian startup founder also highlighted that significant resource constraints mean that large technology companies continue to benefit disproportionately from open resources. To truly localise AI development and address diverse local needs across different regions and contexts, fundamental changes are required, including investment in decentralised compute infrastructure and adopting a frugal AI approach—that means designing and deploying machine learning models that minimise computational costs, energy consumption, and data requirements without sacrificing essential performance. Another key discussion point was the need to simplify the deployment of open source models and enhance their safety by ensuring developers "ship the car with the seatbelts".
These points closely align with the DPGA Secretariat's work. For instance, the DPG Standard mandates that AI systems prioritise safety in development and publicly license their full training data—requirements potentially more manageable for smaller, task-specific models.
The session also featured the launch of a policy brief, “Advancing Open Source AI in India: Recommendations for Governments and Technology Developers”. This brief, which mirrors the efforts of the DPG Standard, encourages developers to provide high-quality, reusable core components. It emphasises avoiding "open washing" through clear documentation, use of safeguards, and a firm commitment to open source AI as the foundation for transparent, local collaboration.
Collaboration for Impact: Climate, Health, and Agriculture
The DPGA Secretariat week was wrapped up by co-hosting a high-level event with the Royal Norwegian Embassy in New Delhi and Norad, focusing on “International Collaboration on AI and Digital Public Goods to Advance Progress on Climate, Health and Agriculture”. The event, which included opening remarks from Norwegian Minister of Digitalisation and Public Governance, Karianne Tung, and the Minister of Management and Innovation in Public Services of Brazil, Esther Dweck, provided practical examples of how AI intersects with digital public goods. This was followed by showcasing the updated requirements of the DPG Standard for AI systems to be recognised as digital public goods, highlighting the importance of data transparency in promoting responsible AI practices.
Representatives from verified DPGs, including DHIS2, Rural Environmental Registry (RER), DiCRA, and ODK showcased how AI can be responsibly integrated into existing solutions. They also demonstrated how their solutions offer trusted data infrastructures essential for AI deployment.
We concluded the discussion by highlighting the value of DPGs as platforms that demonstrate best practices in the era of AI-driven software development. A key point was the necessity of public investment in open datasets to overcome market failures, aggregate markets, and incentivise the private sector to solve social challenges.
“How do we ensure that the digital world grows in ways that are fair, open, and beneficial for all? The answer lies in supporting digital public goods—through our policies, partnerships, and priorities. When digital resources are shared, everyone benefits. And when everyone can participate, innovation becomes not only faster, but also more just and more sustainable.” Karianne Tung, Minister of Digitalisation and Public Governance, Norway
By positioning DPGs as an infrastructure of trust, they offer a value proposition that proprietary solutions cannot match. This is because the DPG Standard ensures that DPGs follow industry best practices, prevent “open-washing,” embed do-no-harm principles by design, and support the attainment of the Sustainable Development Goals. To this end, we were also happy to meet representatives of openFN, Syft and Care | Open Healthcare Network during the summit, which once more underlined the point that DPGs are relevant solutions in developing responsible AI and AI-mediated access to services and information.
Reclaiming AI for the Public Interest
The summit reinforced a powerful truth: an open approach to AI, complemented by participatory governance, guardrails and a deep understanding of the prevalent power imbalances in the AI ecosystem, is the only way to ensure local relevance and agency for the Global Majority.
As Prime Minister Modi noted in his opening remarks of the summit, India believes AI’s benefits must be shared—invoking the principle that "sunlight is the best disinfectant" to make AI safer and more effective. This call for open code and collaborative development aligns deeply with the DPGA’s work. However, to truly reclaim AI as a technology that serves the public interest, we must learn the lessons of open source software by pairing openness with deep, public-sector-led investments in research, safety, and trusted data infrastructures. Open data plays a fundamental role in trustworthy AI, and it was reassuring to hear the recurring call for more open data in AI development-an approach the DPGA Secretariat has championed with the update of the DPG Standard for AI systems last year. By maintaining this high bar and demanding a genuine plurality of models, visions, representations and stakeholders, we can move toward a global AI ecosystem that empowers everyone to build and own AI on their own terms.
The Road Ahead: Our Commitment
The DPGA Secretariat is committed to building the technical and policy foundations for a public-interest AI ecosystem. Our next steps include:
- Strengthening the DPG Standard for AI systems: Conducting a systemic review in the second half of 2026 to assess the open data requirement and continuing our engagement with experts and the community on AI trust and safety best practices.
- Building the DPG4AI Collection: Identifying and onboarding new DPGs that serve responsible AI development, from open data for training and benchmarking to tools and software for packaged, sovereign solutions.
- Making DPGs Future-proof: Providing case studies and insights into the responsible use of AI by established DPGs and positioning DPGs as a platform of best practices.
- Mobilising for Policy Change: Advocating for public sector investment in the open source AI ecosystem to ensure AI is developed with the public interest at its core.
- Supporting Discoverability and Global Collaboration: Feeding into initiatives like the Global AI Impact Commons to ensure everyone walks the talk, and partnering with like-minded organisations to build the public interest AI movement.
