Global AI Dialogue Puts Safety, Inclusion and Oversight at Center of UN Governance Push

As AI For Good opens in Geneva, governments and experts warn that the world’s AI divide is becoming a governance divide.

GENEVA — The United Nations’ first Global Dialogue on AI Governance has put a sharper focus on one of the defining questions facing the international system: whether artificial intelligence will be governed broadly enough, safely enough and inclusively enough to benefit all countries — not only those already leading the technology race.

Held in Geneva alongside the WSIS Forum and the ITU’s AI for Good Global Summit, the Dialogue brought together governments, UN agencies, technical experts, civil society, academia and industry to examine how global AI governance can move from principle to practice. The discussions underscored a growing concern that AI is advancing faster than oversight systems, while many developing countries still lack the infrastructure, regulatory capacity and technical expertise needed to shape the rules that will affect them. 

The Geneva session marks the first of two planned UN Global Dialogue meetings, with a second session scheduled to take place in New York in May 2027. The process is being co-chaired by El Salvador’s Permanent Representative to the UN, Egriselda López, and Estonia’s Permanent Representative, Rein Tammsaar. 

From AI Promise to AI Oversight

Across the opening discussions, speakers emphasized that the governance challenge is no longer theoretical. AI systems are already shaping economies, public services, information environments, education, health care and security. But the systems designed to test, monitor and hold those technologies accountable remain uneven, fragmented and often dependent on the same companies building the tools.

The ITU’s coverage of the Dialogue highlighted concerns that current AI safety systems are frequently developed without sufficient evidence, transparency, accountability or human oversight. Panelists called for safeguards throughout the entire AI lifecycle, rather than only during model training.

UN Secretary-General António Guterres used the Geneva gathering to warn that AI is developing faster than rules can keep up, calling for globally harmonized guardrails and stronger protections for children, including an AI Child Safety Pledge requiring companies to ensure that technologies are safe for children before release.

The New AI Divide

The Dialogue also framed AI inequality as more than a question of access to tools. It is increasingly a question of who has the power to evaluate, regulate and influence the systems that are being deployed globally.

“One of the least-discussed AI divides is the lack of oversight capacity,” Urvashi Aneja, Founding Director of Digital Futures Lab in Goa, India, said during the discussions, according to ITU. 

That framing reflects a broader concern running through the Geneva meetings: countries without advanced computing infrastructure, independent testing institutions, AI safety expertise or strong digital governance systems risk becoming rule-takers in a global AI economy shaped elsewhere.

Reuters reported that Guterres warned of AI power becoming concentrated in a small number of countries and corporations, while developing countries risk being left further behind. The same report noted that the United States holds the overwhelming majority of top AI supercomputing power, followed by China, with much of the rest of the world far behind.

Building a Governance Architecture

The Global Dialogue follows the UN’s broader effort to turn the Global Digital Compact into a more practical framework for AI governance. Its work is also informed by the UN-backed Independent International Scientific Panel on Artificial Intelligence, a 40-expert body created to help governments assess AI opportunities and risks with stronger evidence.

That panel recently warned that rapid and largely unchecked AI development could create catastrophic risks, including misuse for cyberattacks, biological threats, disinformation and autonomous systems that are difficult to control. It also raised concerns that governments often lack timely, independent evidence about frontier AI systems, leaving policymakers dependent on company-provided information.

The Geneva Dialogue did not produce a binding treaty. Instead, it is intended to build shared understanding, identify governance gaps and help shape the next phase of international coordination.

ITU’s Central Role in the Geneva Moment

For ITU, the Dialogue reinforces Geneva’s emerging role as a central hub for AI governance, digital cooperation and development-focused technology policy.

By convening the AI For Good Global Summit alongside the WSIS Forum and the Global Dialogue, ITU has helped create a rare multilateral setting where technical expertise, diplomatic negotiation and development priorities meet in the same week. That alignment matters: AI governance is not only about preventing harm from frontier models, but also about ensuring that AI systems are accessible, trustworthy and useful for countries working to advance health, education, climate resilience, agriculture and public services.

The Geneva meetings also reflect a shift in the AI debate from high-level ethics statements toward implementation: oversight capacity, safety testing, public-interest infrastructure, child protection, inclusion of developing countries and the question of whether global institutions can keep pace with private-sector innovation.

What Comes Next

The next test will be whether the Dialogue can translate Geneva’s concerns into practical cooperation before the New York session in 2027.

For developing countries, the priority will be ensuring that AI governance includes financing, capacity-building and technical support — not just voluntary principles. For regulators, the challenge will be building institutions capable of evaluating powerful AI systems independently. For companies, the message from Geneva was increasingly clear: trust will require more than innovation. It will require accountability.

As the AI For Good Summit continues, the Global Dialogue has given the week a broader political frame. The question is no longer whether AI can support development. It is whether the world can build governance systems strong enough, inclusive enough and fast enough to ensure that it does.