Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Advertisement
PubMed Google Scholar
You have full access to this article via your institution.
The India AI Impact Summit was attended by 20 heads of states and delegates from over 80 countries. Credit: India AI Impact Summit 2026
Eighty-six countries and two international organisations have endorsed the New Delhi Declaration on artificial intelligence (AI), aimed at building an inclusive, human-centric and development-oriented global AI framework.
The consensus is being seen alongside nuanced debates during the India AI Impact Summit (16-21 February 2026) over limitations in sovereignty, access, infrastructure control and the future of global AI governance.
India’s minister for electronics and IT, Ashwini Vaishnaw, framed the declaration as proof of “broad global support” for a human-centric AI vision — one that prioritises access, skilling, and responsible deployment.
The document is expected to be signed at another international summit later this year by countries, including the United States, China, the United Kingdom, the European Union, Israel, Canada, Russia, Saudi Arabia, Singapore, Japan and Australia. The agreement calls for expanding AI benefits to developing economies, strengthening public-interest use cases in healthcare and education, and promoting trust and accountability, Vaishnaw said.
The non-binding declaration will guide national policies, boost cross-border research, and set standards that are consistent, measurable and future-ready, he said.
It also calls for tackling pressing AI challenges — algorithmic bias, cybersecurity risks, workforce disruption, and societal impact. Participating countries have pledged to uphold ethical safeguards, ensure transparency in algorithms, and implement strong monitoring and audit mechanisms, Vaishnaw said.
Balaraman Ravindran at the Centre for Responsible AI at IIT Madras, who chaired one of the working groups at the summit, said the declaration turns attention towards the challenges and risks faced by the global south. “While operational details are thin, as is expected from a multilateral declaration, the agreement has consensus from so many countries, including the two AI superpowers.”
“That is an acknowledgement of the growing need to focus on immediate challenges to beneficial AI adoption across world,” he said.
During discussions at the summit, AI experts from several countries noted that this agreement on principles must look at tackling structural limitations.
Earlier at the summit, leaders from small and developing countries stressed that AI access remains deeply unequal.
Referencing Mauritius and other small economies, the Prime Minister of Mauritius, Navin Ramgoolam, warned that countries outside major blocs do not enjoy the same financing tools such as concessionary loans or subsidies and have limited access to capital for research and development. Without external partnerships, they said, smaller states “do not have the capacity to invest in the R and D that is required.”
The Vice-President of Seychelles, Sebastien Pillay, said small states may lack oil or minerals but possess “human capital” — and want AI to strengthen government efficiency, economic diversification, food security, and biosecurity. Realising those ambitions, however, requires sustained technology transfer and legal readiness, not just diplomatic language, he said.
Pillay echoed a core promise of the Delhi Declaration that for vulnerable economies, the democratization of AI must translate into infrastructure, finance, and technical cooperation.
The American delegate Michael Kratsios, Director of the White House Office of Science and Technology Policy, cautioned against framing AI as a binary between “haves and haves-nots,” arguing that such framing risks obscuring the central task of enabling governments to deploy the “best AI technology” strategically.
Kratsios signalled clear resistance to centralized global oversight, emphasizing instead “sovereign AI capability” and the rejection of global AI governance. AI, he suggested, should advance through trade and partnership rather than supranational regulatory structures. American AI, he said, is “open for business” and would prioritise “trade over aid”.
While the Delhi declaration stresses cooperative guardrails, Washington’s emphasis remained on national capacity, exports, and strategic partnerships. Kratsios illustrated these limitations in achieving a fully harmonized global framework.
Serbian President Aleksandar Vučić sharpened the geopolitical dimension by arguing that AI is rapidly becoming infrastructure, and that infrastructure is inherently political.
Invoking Albert Einstein’s warning that technology can outpace wisdom, Vučić wondered whether political systems can keep pace with AI’s acceleration. He warned of an “unprecedented concentration of technological power” and posed a blunt question: will a small number of actors set the rules for everyone else?
“Sovereignty in the 21st Century,” he said, includes the ability to control data, regulate algorithms, and develop domestic expertise. Without that capacity, sovereignty risks becoming merely formal.
These concerns resonated beyond Europe. As large language models, compute clusters, and semiconductor supply chains consolidate within a handful of economies and corporations, speakers at the leaders’ plenary at the summit, chaired by India’s Prime Minister Narendra Modi, reflected on the politics of AI increasingly centering on who owns infrastructure and who defines standards — a key limitation in realizing global equity.
Slovakian President Peter Pellegrini echoed this infrastructure focus, stating that AI “must not stay in the hands of few.” Democratization, he argued, requires real access to skills, tools, and fair conditions for innovation.
This sentiment was echoed by heads of major AI company representatives at the summit, including Sam Altman of OpenAI, Sundar Pichai of Alphabet/Google, Demis Hassabis of Google DeepMind, Brad Smith of Microsoft and Dario Amodei of Anthropic.
“Computing power is the new infrastructure,” Pellegrini said.
AI requires energy-intensive data centres, advanced Graphic Processing Units (GPUs), and stable jurisdictional frameworks. Slovakia highlighted its low-carbon energy base and investments in domestic supercomputing, framing compute sovereignty as both an economic and strategic imperative.
At the same time, Pellegrini insisted that responsibility must “always remain with a human being”, reinforcing a central theme of the Delhi Declaration – human oversight as a baseline principle.
The New Delhi Declaration articulates shared commitments to responsible, inclusive AI and greater global cooperation. It emphasizes expanding access for developing countries, strengthening digital public infrastructure, and ensuring trust through transparency and safeguards.
The declaration non-binding and does not establish enforcement mechanisms. It does not directly confront concentration of compute among a handful of corporations. Nor does it reconcile differing philosophies between export-driven national sovereignty and multilateral regulatory frameworks.
In effect, it codifies principles but leaves implementation contested, revealing the limitations of multilateral agreements in a rapidly evolving technological landscape.
doi: https://doi.org/10.1038/d44151-026-00036-6
Reprints and permissions
You have full access to this article via your institution.
An essential round-up of science news, opinion and analysis, delivered to your inbox every weekday.
Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.
Nature India (Nat India)
ISSN 1755-3180 (online)
© 2026 Springer Nature Limited
Leave a Reply