United Nations Secretary General António Guterres is calling for stronger international coordination on artificial intelligence as governments and technology companies race to develop increasingly powerful systems. Speaking ahead of the 2026 United Nations General Assembly high level week, Guterres warned that AI development is moving faster than the world’s ability to understand and govern its consequences. His appeal places global AI safety, human oversight, transparency, and international cooperation at the center of an increasingly urgent policy debate.
Guterres Warns That AI Governance Cannot Remain Fragmented
Guterres has argued that no country can fully manage the opportunities and risks created by advanced artificial intelligence on its own. His latest remarks reflect a broader United Nations effort to establish common principles for AI governance at a time when major technology powers are pursuing different regulatory approaches.
On September 16, Guterres identified runaway AI as one of three existential threats facing the international community, alongside the climate crisis and deepening inequalities. He said governments need international cooperation because AI development is progressing faster than humanity’s understanding of its potential consequences. The Secretary General also warned against isolated voluntary pauses, arguing that limited participation could leave countries competing against one another on AI safety.
The United Nations has already established a Global Dialogue on AI Governance, giving governments and other stakeholders a forum for discussing common approaches. The first session took place in Geneva in July 2026, with participation designed to include countries from across the international system. The UN says the initiative grew from commitments made through the Global Digital Compact and the Pact for the Future.
Why Common AI Safety Rules Matter
The central concern is not simply whether AI should be regulated. It is how countries can establish rules that work across borders while allowing legitimate research and technological development to continue.
An AI system can be developed in one country, trained using information gathered internationally, operated through infrastructure located somewhere else, and used by people on several continents. A safety problem therefore does not necessarily remain within the jurisdiction where the system was created.
Guterres has called for common approaches to testing AI systems, evaluating risks, and assigning responsibility. The goal is to reduce the possibility of a fragmented regulatory environment in which companies must follow completely different technical requirements in different markets.
For businesses, shared standards could make international compliance more predictable. For governments, common testing methods could provide a clearer basis for judging whether increasingly powerful systems are safe enough for deployment. For consumers, consistent safeguards could make it easier to understand what protections apply when AI services cross national borders.
The UN Wants Human Responsibility to Remain at the Center
One of the most significant themes in the UN position is human accountability. Guterres has repeatedly argued that artificial intelligence should assist people rather than remove human responsibility from important decisions.
That principle becomes especially significant when AI is used in areas such as healthcare, criminal justice, policing, education, employment, and financial services. An automated system may process information quickly, but people can still be affected by inaccurate data, flawed models, discriminatory outcomes, or decisions that are difficult to challenge.
The Secretary General has argued that machines may inform high stakes decisions, but humans must remain responsible for making and reviewing them. That approach would require organizations using AI to provide meaningful oversight rather than treating an algorithm as the final authority.
Children Are Becoming a Major Focus of AI Safety Efforts
Children are another major concern in the United Nations approach to AI governance. Guterres has warned that young people are already interacting with AI systems in education, entertainment, communication, and personal conversations without universal standards requiring child specific safety testing.
At the July Global Dialogue on AI Governance, he called for an AI Child Safety Pledge built around several principles. These include child specific safety testing, independent oversight, stronger protections against the generation and distribution of child sexual abuse material, and systems that connect children experiencing serious distress with human support.
The concern is particularly personal for families. A child using an AI chatbot may not understand the difference between a generated response and advice from a qualified human being. A system that sounds confident can create a powerful impression of authority even when its information is incomplete or wrong.
That is why child safety raises questions that go beyond traditional content moderation. Regulators and developers must consider how systems behave, how they respond to vulnerable users, how harmful material is detected, and whether human intervention is available when automated safeguards fail.
AI Safety Is Also Becoming a Security Issue
The international discussion has expanded beyond civilian applications. Guterres has warned that AI technologies and advanced computing hardware developed for civilian purposes can also be used in military settings.
His strongest concerns involve autonomous weapons capable of selecting and engaging targets without meaningful human control. He has argued that decisions involving the taking of human life should remain under human judgment and called for lethal autonomous weapons to be prohibited through international law.
This issue presents one of the most difficult challenges for global AI governance because military technology is closely connected to national security. Governments may be reluctant to disclose capabilities, testing programs, or strategic objectives. At the same time, rapid advances in autonomous systems could create pressure for competing states to develop similar technologies.
International rules would therefore need to address both technological capabilities and the circumstances in which those capabilities may be used. That requires sustained diplomatic negotiations rather than isolated statements from individual governments.
AI Governance Must Include Developing Countries
Another major issue raised by the United Nations is the unequal distribution of AI infrastructure. The most advanced systems require large amounts of computing power, specialized chips, data, electricity, and technical expertise. These resources remain concentrated among a relatively small number of companies and countries.
Guterres has warned that the world could move from a digital divide to an AI divide if developing countries lack access to computing capacity, skills, data resources, and technical knowledge.
The UN has responded by supporting an international network focused on AI capacity building. Guterres has also proposed a Global Fund for AI intended to help developing countries build skills, access computing resources, and develop stronger local AI capabilities.
This issue matters because regulation without capacity can create another form of inequality. Countries may be expected to enforce sophisticated AI rules while lacking the technical resources needed to test systems, investigate violations, or develop their own alternatives.
The Environmental Cost of AI Is Entering the Debate
AI safety discussions are also expanding to include energy, water, and land use. Large data centers require substantial electricity and cooling resources, and the expansion of AI infrastructure could increase pressure on energy systems in regions hosting these facilities.
Guterres has called for greater environmental transparency from major AI companies, including reporting on carbon emissions, water use, and land impacts. He has also encouraged companies to move toward renewable energy for data center operations.
The environmental question adds another layer to the global governance debate. A country may benefit economically from hosting AI infrastructure while local communities bear some of the costs through increased demand for electricity, water, and land. Transparent reporting could help governments and communities evaluate those tradeoffs more clearly.
Science Is Being Used to Build a Shared Evidence Base
The United Nations has also created an Independent International Scientific Panel on Artificial Intelligence. The panel is intended to provide governments with independent scientific assessments of AI’s capabilities, risks, and effects across economies and societies.
The UN describes the panel as a way to reduce the knowledge gap between countries and give international negotiations a shared evidence base. This matters because governments can reach very different conclusions when they rely on competing technical assessments.
A scientific assessment cannot determine political choices on its own, but it can help policymakers distinguish measurable risks from speculation. That distinction will become increasingly important as governments debate regulations covering frontier AI systems, automated decision making, data governance, cybersecurity, and emerging military applications.
The United Nations has published further information about its global AI governance work, including the development of international dialogue and scientific mechanisms intended to support cooperation among governments and other stakeholders.
Technology Companies Will Face Greater Pressure for Accountability
Global AI rules would also place greater responsibility on companies developing and deploying advanced systems. Technology firms have invested heavily in increasingly capable models, while governments are still working to determine how existing laws apply to technologies that can generate text, images, audio, software, and other forms of content at enormous scale.
Guterres has argued that companies should accept responsibility proportional to the power of their systems. That could mean stronger pre deployment testing, independent evaluations, transparent risk reporting, safeguards for vulnerable users, and clearer mechanisms for correcting harmful outcomes.
For responsible developers, predictable rules could also provide benefits. A common international baseline could reduce uncertainty over which safety practices are expected when an AI product reaches users in multiple countries.
The Challenge Is Finding Rules That Work Across Borders
Creating a unified global AI framework will not be simple. Countries have different economic priorities, legal systems, national security interests, technology industries, and views about privacy and individual rights.
The United States, China, European countries, developing economies, and other governments may also place different levels of importance on innovation, national competitiveness, security, and regulation. Reaching agreement on basic safety principles could therefore be easier than agreeing on every detail of AI policy.
That distinction may prove important. Global governance does not necessarily require every country to adopt identical laws. It could instead involve common safety principles, compatible technical standards, shared testing methods, international reporting mechanisms, and agreed limits around particularly dangerous applications.
What Comes Next for Global AI Regulation
The immediate challenge is turning broad principles into practical rules. Governments will need to determine which AI systems require the strictest oversight, what constitutes an unacceptable level of risk, who should conduct safety evaluations, and what consequences should apply when companies fail to meet established standards.
The next phase will also require participation from technology companies, scientists, civil society groups, educators, and communities that are already experiencing the effects of AI. A framework designed only by governments or technology companies would leave important perspectives outside the discussion.
The UN’s Global Dialogue on AI Governance is scheduled to continue in New York in May 2027. Its development reflects a broader recognition that AI governance cannot be treated as a single national policy question.
For people using AI every day, the debate may sound distant, but its consequences are increasingly practical. The rules eventually established could influence how AI systems handle personal information, how automated decisions can be challenged, what protections children receive, how companies disclose risks, and who remains accountable when technology causes harm.
A Global AI Framework Could Shape the Next Decade
Guterres’s latest warning arrives at a moment when artificial intelligence is moving from an emerging technology into a basic part of economic and social infrastructure. The central question is no longer whether governments should pay attention to AI governance. It is how quickly they can develop workable safeguards while the technology continues to advance.
The United Nations is advocating a model based on international cooperation, scientific evidence, human oversight, transparency, human rights, and broader access to AI capabilities. Whether major powers and technology companies can translate those principles into binding and practical arrangements remains an open question.
What is becoming clearer is that national borders do not provide a complete solution to AI risks. A model released in one market can reach users worldwide within hours. A deepfake can cross continents before authorities can respond. An automated system can influence decisions far from the location where it was built.
That reality gives the global governance debate a distinctly human dimension. The technology may be built from algorithms, servers, chips, and data, but the consequences are experienced by people in classrooms, hospitals, workplaces, homes, and communities. The international rules developed now will help determine whether artificial intelligence becomes a source of wider opportunity while remaining subject to meaningful human responsibility and safety.
