New Delhi Declaration: 89 Countries Discuss Pathways for AI Governance

On February 20, the five-day India AI Impact Summit concluded in New Delhi. On the same day, India released the "AI Impact Summit Declaration," providing a voluntary, non-binding framework document for global artificial intelligence cooperation. As of the 21st, 86 countries, including major economies such as the United States, the United Kingdom, China, and Germany, along with three international organizations, have signed this declaration. This document, born in South Asia, shifts the focus of global AI governance toward development and inclusiveness.

Declaration Framework: Development Supported by "Seven Wheels"

The philosophical foundation of this "New Delhi Declaration" originates from the Sanskrit maxim Sarvajan Hitaya, Sarvajan Sukhaya, which means for the welfare of all, for the happiness of all. On the 21st, Indian Minister of Electronics and Information Technology Ashwini Vaishnaw stated that the declaration aims to advance Prime Minister Modi's vision of human-centric artificial intelligence. Unlike previous international initiatives that focused primarily on safety and governance, this declaration places greater emphasis on development and is structured around seven pillars, referred to as Chakras (meaning wheels).

The seven pillars include: human capital development, social empowerment, trustworthy AI systems, energy efficiency, scientific AI, democratization of AI resources, and AI that promotes economic growth and social welfare. Among these, the democratization of AI resources holds a key position. The declaration specifically mentions the "Charter for Democratic Diffusion of Artificial Intelligence," which aims to enhance the accessibility of computing power, data, and tools in developing countries while respecting national laws and sovereignty. India hopes that through this framework, it can address the common challenges faced by Global South countries in AI development, such as insufficient computing power, lack of data, and talent shortages. Vaishnau revealed that during this summit alone, parties secured over $250 billion in investment commitments for AI infrastructure, indicating that resource allocation is beginning to shift from initiatives to action.

The declaration also emphasizes the value of open-source artificial intelligence and scalable use cases, proposing to promote the replication of successful AI applications across different regions through the voluntary platform of the Global AI Impact Commons. In terms of trustworthiness and security, the declaration supports industry-led voluntary measures and technical safeguards, and mentions the collaborative toolkit known as the Trustworthy AI Commons. Analysis suggests that this structure attempts to balance innovation with public interest. However, as the declaration is non-binding, its actual effectiveness will depend on the subsequent independent actions of countries and international cooperation.

Signatory Map: Common Demands Amid Differences

Eighty-nine signatories form a complex geopolitical landscape. Members of the G7, such as the United States and the United Kingdom, are listed alongside Global South countries including China, Egypt, and Indonesia. Despite differences in technological development levels, regulatory approaches, and strategic interests, the act of signing itself reflects that global AI governance is evolving toward a cooperative dynamic that transcends traditional blocs.

For developed economies, participating in the New Delhi Declaration complements existing governance processes. They may view this declaration as a channel to engage with emerging markets and influence the development pathways of artificial intelligence in the Global South. For China, the declaration's emphasis on respecting national sovereignty and technology serving development aligns with its consistent stance, while also demonstrating a cooperative posture within multilateral frameworks.

The deeper reason lies in the fact that global challenges brought by artificial intelligence, such as the skills gap, employment impacts, and energy consumption, require a minimum level of international cooperation. The declaration's contents, including the establishment of an international network of scientific AI institutions and the development of guiding principles for AI workforce retraining, precisely address these cross-border issues. As the host country, India successfully played a bridging role: it showcased its own technological ambitions—launching three indigenous AI models during the summit—and strengthened its diplomatic image as a spokesperson for the Global South. In his radio program on the 22nd, Indian Prime Minister Narendra Modi said, "Some people fear artificial intelligence; India sees the future in it." This statement highlights the optimistic tone India aims to convey.

From Vision to Practice: Opportunities and Challenges Coexist

The summit also served as a grand exhibition. According to Vaishnau, over 500,000 people visited the related exhibitions. Modi highlighted two cases that have garnered international attention, reflecting India's vision for the application of artificial intelligence.

At the booth of Indian dairy brand Amul, the first case demonstrates the application of artificial intelligence in dairy cow health monitoring and milk production management. This case highlights the potential of AI to enhance productivity in traditional agriculture. The second case focuses on cultural heritage preservation, using AI to improve the image quality of ancient manuscripts, converting them into readable formats, and then translating them into multiple languages through AI. Modi mentioned that leaders from various countries showed great interest in understanding ancient Indian texts through this approach.

These specific applications resonate with the theme of social welfare in the declaration, illustrating that artificial intelligence can not only be applied to cutting-edge technology but also address developmental issues such as food security and cultural heritage. However, there remains a gap between vision and reality. The declaration is voluntary, and its seven pillars lack mandatory timelines and funding mechanisms. Scaling up the AI pilot from Gujarat to smallholder farming communities in Africa or Southeast Asia still requires overcoming a series of challenges, including infrastructure, digital literacy, and localized adaptation. History shows that technological dividends do not spread automatically.

Soft framework, practical impact

The New Delhi Declaration emerged at a time when the global governance landscape for artificial intelligence is rapidly evolving. The European Union's Artificial Intelligence Act has entered the implementation phase, the United Nations is advancing the establishment of a scientific advisory body for AI, and the United States is also promoting standard-setting through executive orders and industry alliances. Against this backdrop, this declaration, centered on cooperation and development, introduces a new variable into the global dialogue.

It may bring about three impacts. First, it formally elevates issues such as AI democratization and access to computing resources to the global agenda, prompting developed countries to respond to the demands of developing nations. Second, it may give rise to new multilateral projects; for example, if the Global AI Impact Commons is launched, it will become the first platform dedicated to cross-border replication of AI social application solutions. Third, it provides emerging countries with a set of referential policy discourse and a principled framework, which may influence the formulation of their domestic AI strategies.

However, the limitations of the declaration are also evident. As a soft law document, it struggles to address the most challenging issues in AI governance, such as the regulation of autonomous weapon systems, international auditing of advanced AI, and binding rules for cross-border data flows. Moreover, despite the large number of signatories, the role of core technology companies in the global AI field within the declaration is primarily defined through voluntary industry measures, resulting in limited enforceability.

The meeting has concluded, awaiting the implementation of the declaration. Behind the 89 signatories lie diverse interests, capabilities, and expectations. This document will not become the United Nations Charter of the artificial intelligence era, but it opens a window for global dialogue—which has been overly focused on risk control—to shift attention toward development and inclusion. Its historical significance will not be determined by the designs on paper, but by the coming years: whether computing resources truly flow to laboratories in Nairobi or Dhaka, whether affordable AI tools help Southeast Asian farmers cope with climate change, and whether the benefits of AI growth can be shared more equitably. This journey has only just begun.