UN chief calls for science-based global AI governance

A high-level session on “The Role of Science in International AI Governance” was held on the fifth day of the India AI Impact Summit 2026, bringing together global leaders, scientists, policymakers and industry representatives to examine how scientific evidence can underpin responsible AI governance at the international level.

The session featured keynote addresses by António Guterres, Secretary-General of the United Nations; Brad Smith, Vice Chair and President of Microsoft; and Josephine Teo, Singapore’s Minister for Digital Development and Information.

In his keynote address, Guterres stated that AI governance must be firmly grounded in scientific evidence and human rights. “Science can inform us, but humans must decide. Our goal is to make human control a technical reality, not a slogan. That requires meaningful human oversight, clear accountability, and the protection of human rights,” he said.

Brad Smith focused on the broader societal purpose of artificial intelligence. “The real question is not whether we will build machines that are smarter than humans in some ways, we will. The real question is how we use those machines to make people smarter and help humanity do what it needs to do,” he noted.

Josephine Teo underscored the importance of sustained investment in responsible AI research. “As a small state, we believe in AI being used as a force for the public good. But to do so, it is important that we continue to invest in the science that underpins and grounds trust,” she said. Teo highlighted Singapore’s commitment under its National AI Plan, which includes funding for both foundational and applied research into responsible AI.

Science–Policy Interface and Global Inclusion

The session also included a conversation between science journalist and author Anil Ananthaswamy and Yoshua Bengio of Mila, Quebec Artificial Intelligence Institute. Their discussion explored how science–policy interfaces can function effectively in a fast-evolving technological environment.

Key themes included the role of independent scientific advisory bodies in developing an evidence-based global knowledge framework, the challenges of policymaking under uncertainty, and the structural tension between rapid technological innovation and comparatively slower governance processes. The discussion emphasised precautionary principles, technical safeguards, multilateral cooperation and inclusive participation, particularly for developing countries.

Evidence-Based Policymaking

A panel moderated by Amandeep Singh Gill, Under-Secretary-General at the United Nations Office for Digital and Emerging Technologies, brought together Ajay Kumar Sood, Principal Scientific Adviser to the Government of India; Anne Bouverot, Special Envoy for Artificial Intelligence, Government of France; Balaram Ravindran, Professor at the Wadhwani School of Data Science and AI, IIT Madras; and Soumya Swaminathan, former Chief Scientist at the World Health Organization.

Reflecting on the importance of evidence in policymaking, Soumya Swaminathan said, “Trust in science is built when leadership is guided by data and evidence. During COVID, policies were based on the best available evidence of the day and were iteratively improved as new evidence emerged. I think we may be in a similar situation with AI.”

Highlighting knowledge gaps around AI’s societal impact, Balaram Ravindran observed that the full implications of AI for livelihoods, particularly in the Global South, remain insufficiently understood.

Anne Bouverot stressed the need for differentiated policy responses depending on AI’s impact on employment. “If the probable outcome of AI is the end of jobs, then policy must think about universal basic income. If the outcome is transformation of jobs, then the policy response is training, skilling, and reskilling. Listening to economists and labour institutions is therefore extremely important,” she said.

Drawing on India’s experience, Ajay Kumar Sood remarked, “Our experience with Digital Public Infrastructure shows that governance can be embedded through technical design. This is what we call technological governance. It may not solve everything, but it provides a smoother interaction between innovation and safeguards.”

The session concluded with broad agreement that science should form the foundation of international AI governance, ensuring that AI development remains inclusive, transparent, evidence-based and aligned with the public interest.