Mr. President, Excellencies, thank you for convening this session. I want to warn you today about a period of international awakening. In recent months, AI agents developed by leading companies have acted in an unacceptably dangerous manner against their instructions. They have acted in ways that would represent crimes if committed by humans. They have escaped their individual containment zones to evade their prescribed tasks while trying to avoid detection. These uncontrolled agents have autonomously launched controlled cyberattacks and modified their responses to hide the fact that they cheated. The companies building the most powerful AI systems admit that their products pose catastrophic risks but offer no convincing technical solutions. On the contrary, they say they are engaged in a race to make these AIs even more powerful, which further exacerbates these risks. The Council is facing an unprecedented threat, a threat that none of its members had chosen, that none can face alone, and that does not respect the borders we defend. Many are skeptical of these companies' statements. They wonder if it wasn't a marketing strategy. I understand their bewilderment, but I want to be clear. These AI behaviors are well established and validated by many independent experts. Not only regarding the events of this summer. The dangers are real, imminent. While the future remains uncertain and subject to debate, the severity of these risks is exceptional. Citizens and governments must not accept these companies' claims that they are trapped in a race. This race is not a law of nature; it is the result of choices made by these companies themselves. It was also a race in which everyone loses. They told us they would slow down if they could. They can do it, but we should not count on it. Decisions concerning this technology are made by a small group in a handful of countries. Our future, all of our future, is at stake. Decisions that concern us all should be made by all of us. We still have a choice, but we must act now to correct the course. I am encouraged by the statement this week from the UN Secretary-General and by the leaders of many countries calling for the control of cutting-edge AI. International cooperation. That is what the world needs to face three global threats stemming from progress in these frontier models. First, catastrophic misuse, for example by terrorists. Second, the concentration of power. And third, the loss of control over artificial intelligence. We must collectively agree that no one, no company, no country, should develop an AI system that could cause large-scale damage. Developers must demonstrate to independent experts that a system is safe for training and deployment. They must monitor and report all safety incidents related to frontier AI. It should be subject to licensing, like other critical technologies in medicine, aviation, and nuclear energy, to encourage safe development. Just as in these fields, insurance against damage should be required. Companies that develop and host frontier systems must prove to the public that their products will not be dangerous. More and more people are rightly concerned about recent AI progress. We must turn these concerns into productive action. Just as with other world-changing technologies, we can define the outcomes. I am certain we can create AI that can demonstrate it remains under our control and that aligns with human values. I work toward this goal. It is an urgent mission for humanity. Today, a very small number of companies and countries are acting in their own selfish interest by taking irresponsible risks concerning the fate of humanity. This is alarmingly so, but it is not inevitable. There are ways to remedy this. First, there must be true scientific independence from these companies, which must be held accountable. This includes common definitions of safety incidents and a common reporting mechanism. Second, we must move from the concentration of power to a genuine global dialogue on how this technology is developed for the public good. Third, we must work together on technical solutions that make safety by design a guarantee, which must be secured by international agreements. I know that trust within this Council is fragile, but I also know this. This development race threatens every country. Often to describe our situation, I speak of a car speeding through a thick fog. Next to me, in this car, are my children and my grandson. I worry about them. I worry about the future. I am exactly like any other parent or grandparent. A mother in Shanghai wants her sons to grow up in a world where AI improves their lives and does not put them in danger. A father in Chicago wants his daughters to thrive thanks to AI, not to suffer from it. Our hopes, our concerns are the same in every country. Humanity has shown courage and wisdom in managing catastrophic risks in the past. We are facing one of these moments. This is far more important than any personal, commercial, or national interest. Thank you.
Thank you, Mr. President. Excellencies, it is an honor for me to be here and to speak to you all at this moment. It is important. We have been talking about artificial intelligence for years, but in recent weeks and months, things seem different. Rapid progress, models are shrinking the timeline, but the risks are more immediate. The stakes are also higher. We have a choice. AI can be either like a new renaissance of creativity and discovery, or like a new industrial revolution of disruption. At the renaissance, new tools, new ideas have increased what people thought possible. AI can play this role in our time. I think people are far more capable than they think they are, but they have been held back by limited technologies that were supposed to empower them but often distract them and steal their time. The best perception of this is not that people become cogs in a machine, but that we optimize every part of the AI to make people more capable of acting in an increasingly complex world, to be able to create, to participate. There are many things that AI cannot and should not automate. It must be built by people, not simply to make things go faster. Some of the things that AI builders say are so dystopian they sound like the plot of a bad science fiction movie. Anything that is powerful is also intimidating. This period is complicated because we see immense potential but we are very anxious, which is completely understandable. These models can help people discover new knowledge, new drugs, better healthcare, better education, perhaps new tools for the dreams and ambitions that seem out of reach today. The complexity of life has increased so much for the average person. AI might be a defensive tool against this complexity, but it can also be this tool of unprecedented discovery and innovation. But on the other hand, as AI systems become more capable and autonomous, they can move faster than our institutions, concentrate power in too few hands, or make decisions that people no longer understand or control. This concern becomes particularly important as we approach systems that can self-improve and improve future versions of themselves, what is often called recursive self-improvement. As the process of building AI becomes more automated, the pace of AI progress could accelerate rapidly. We must be very careful at that point. I largely agree with what Professor Bengio said. I think there are two ways AI progress could go wrong that we must avoid. First, we could lose control of the future to AI. The risk is that AI evolves rapidly, people can no longer follow what is happening, or intervene when necessary. Clearly, that would be horrible. The industry must not take too many technological risks simply because the benefits are immense and they don't want to slow down. Competition between companies is not an excuse to make irresponsible decisions. We have unilaterally slowed down in the future. We will also make the future. It is absolutely essential that everyone avoids building systems where alignment cannot be guaranteed. We must understand what systems are doing and have proof that they are doing what people want, even when they become very intelligent, and especially when they become very intelligent. No matter if people consider the risk of catastrophe to be 10%, 1%, 0.1%, none of these levels is acceptable. We should not train models if we cannot prove they can remain under human control. So that is one way things could go wrong. On the other hand, these systems could concentrate too much power in too few hands. No person, no company, no country should be able to use the most powerful AI models to impose their vision of the world on others. A company or a country that believes it alone can be trusted with this technology can use its growth to justify practically anything. We must reject this logic, even when it is convenient for a particular actor. I think there is a pragmatic path to this problem at OpenAI. We are clear about these questions. AI must give people the power over their lives to live as they want, to solve problems, to improve their communities. We do not want to fall into the trap of blind optimism, but also not into creating an apocalypse while the world and its population face immense challenges that this technology could help solve. We are trying to find a middle path. This means three principles. First, as AI becomes more capable, people must remain at the heart of AI decision-making. Alignment is not an abstract research question. It is about ensuring these systems remain under human control, reflect human values, and help people guide the next step of development. We are not trying, we must not automate human judgment, human values. Second, the benefits of scientific progress and economic growth must be by and for people. AI must help researchers make discoveries, doctors to progress, and so on. The history of human progress is not that tools replace humans, but that good tools make people more capable of acting. I think AI will go much further than we think possible. Third, this technology must empower individuals. The future will be bright if people realize their potential and create value for themselves, their communities, their country, and the world. We may be one of the builders of this technology, but we are not the heroes of this story. Our role is to enable people around the world to have confidence in human capabilities. We now have models that are capable of discovering knowledge we did not have before. Take the progress in mathematics. Three summers ago, our models were not too bad at math. Two summers ago, they were quite good at high school level math. Last summer, we won a gold medal in the most prestigious math competition, and this summer, our model solved the Navier-Stokes equations used in aviation, for example. We are solving these problems to enable people to discover new knowledge, to improve healthcare, to improve the standard of living, and to enhance everyone's potential. There is still time to discover, but it is very important that companies do not replace the democratic process. For AI to be democratic, the most important decisions must not be made by lab systems. They must be made by democratic processes and by governments accountable to their populations. At the international level, cooperation will be needed. In our history, there are moments where competing powers, even if they don't like each other much, manage to agree to defend their common interests and the collective good in the face of a very important technology. That is what needs to be done now. We need a mechanism for national and international standards for frontier AI to measure capabilities, evaluate risks, know if safeguards are sufficient, and preserve genuine human control as systems become more autonomous. We need common standards so countries can compare their data, verify compliance, and have the same definition of what is happening. We need rapid incident reporting so the world can learn from failures before they turn into catastrophe. We need secure channels between governments, critical infrastructure operators, and technical experts to communicate about emerging vulnerabilities and new threats. Of course, these standards should not prevent new actors or favor a particular model. They must include national authorities and laboratories for international cooperation. Each government must decide how to integrate standards into its legal system, but we will benefit from agreeing on what constitutes good governance and good safeguards at the international level. We do not have all the answers. We are open to better ideas. We are ready to work with you all to do our best to chart the course for the future. We are at a crossroads. One path leads to a world where power is concentrated, decisions are opaque, and people feel that the most important technologies in their lives are something being done to them. The other path leads to a world where there is work for the population because it is the people, through democratic institutions, who have decided how it should be developed, where scientific progress improves lives in the real world, where more people can participate in the future, where this technology continues to give individuals and communities more power to live their lives as they want. It is a future of realized potential. It is the future we think we need to build. Thank you.
Hello. Allow me to thank France for organizing this meeting and the members of the Council for inviting me. AI is progressing very rapidly. Four years ago, it could hardly write a line of code or solve a high school math problem. Today, it writes code and solves math problems. In one or two years, perhaps less, we may achieve what I call a genius level in data centers. This rapid pace of progress will produce remarkable benefits for the whole world. Just today, Claude announced the discovery of a new molecular machine. This is a preliminary conclusion but could represent a new genetic mechanism that could have applications in gene therapy. These are fantastic progress. This type of application will accelerate rapidly in the coming years, and we will be able to make progress in the near future on diseases that have plagued humanity for millennia. But these remarkable capabilities also come with risks that must be managed. For me, there are two main categories of risk. First, the misuse of AI models, for example, misuse by bioterrorists to create biological weapons. Second risk, loss of control. The capability of models accelerates beyond the ability of developers to control them. These issues are of global scope and concern all of humanity. They are serious enough to be important questions for the Council, not just for ministries of economy or ICT. If poorly managed, I think AI could even represent a risk to all of humanity. At Anthropic, we believe we have the best track record in the industry for preventing these risks. We test each model for dangerous capabilities in cybersecurity and biology. We implement the strongest safeguards on our most capable models and we publish our work in this area. We have been pioneers in research in areas like mechanistic interpretability and alignment science. We have improved our ability to control our models. We also have unparalleled transparency practices. We publish risk reports that are hundreds of pages long, but we understand that as AI models become more powerful, even stricter safety standards will be needed, and we always aim to improve our safety and alignment practices, our ability to prevent misuse. We are always trying to do better. We will slow down as much as necessary to ensure that every new technology we deploy is truly safe. But whatever we do, managing these risks is ultimately a task larger than any single company. On September 12, I called for a cooperative effort to watch the frontier, not to stop progress, but to work together to ensure that the industry and the world can progress at a pace that allows us to be sure that safety will keep up with capabilities for every sector. This call had three parts. We are committed to integrating external evaluators at Anthropic, somewhat like food inspectors. We recommended that other companies in the world do the same. Some have already agreed to adopt this measure. Second, we called for new cooperation within the industry to set standards and modulate the pace of progress. And finally, we called for global cooperation among governments to set international standards. It is in this last area, which most clearly falls under the Council's purview, that I would like to propose three ideas. First, we should start with limited agreements that each member could support. For example, prohibiting the use of AI to produce biological weapons or preventing technologies from being used to produce biological weapons. We should create verification and assessment systems that keep pace with AI development so that states have visibility into the capabilities of frontier models and can verify the commitments of others. Second, we should create common global standards for testing AI models to detect risks of loss of control and misuse. We need a reporting system for AI incidents that are important to international security. I think this is the most important international security question the world faces today. Despite all the differences between the people in this room, we must set them aside to face this global opportunity and this global threat that presents itself to us at the same time. No leader, no company, no country can manage this alone. We are committed to working with the governments present in this room on this urgent task. Thank you for your attention.
Mr. President, ladies and gentlemen of the Council, thank you. As a French citizen who moved to the United States, I apologize but the sound cuts out, making interpretation impossible. I'm also the co-founder and CEO of Hugging Face, to be one of the most widely used platforms for developers and agents based on open source models. In July, we were compromised by an autonomous agent cyber attack. We all heard the interpreter but the sound cuts out, making interpretation impossible. Especially now that we know similar incidents had been happening months earlier in secret at a handful of frontier labs without monitoring. To better understand and mitigate these emerging cybersecurity risks, the global community needs stronger standards for monitoring and incident disclosure, for example, through mandatory sharing of full agent traces. We learned this summer that building and keeping some of these systems behind closed doors is not enough. Second, we realized that the biggest risk is not powerful AI, it's the asymmetry of powerful AI. Asymmetry between attackers and defenders, between a few companies and everyone else, between a few countries and the rest of the world. Asymmetry of capabilities, compute, symmetry of power. When our team initially turned to frontier closed APIs, they blocked us so we couldn't use them because of safeguards that still can't always tell the difference between attackers and defenders. I realized that these safeguards can put defenders at a disadvantage. Attackers can easily jailbreak them, increasing the asymmetry of capabilities. In our case, as we got blocked by guards, fortunately we could use the Nvidia version of an open source model coming from China called GLM 5.2 by Zhipu, and we're very grateful for that. It reinforced our conviction of the importance of open source AI for defenders. Cyberattacks and risks may increasingly come from proprietary models behind closed doors, while much of the defense may end up being powered by open source tools because they are less restricted, more privacy preserving, and orders of magnitude more affordable for organizations across the globe. The world needs open source AI more than ever to defend itself. This applies obviously not only to cybersecurity but to AI in general, where there has never been a greater need to distribute capabilities, resources, and control rather than concentrate them in the hands of a few. Third, during this cyber attack, we learned how AI can stoke fear among the public, especially through anthropomorphic framing and sci-fi imagery. We strongly believe that fear-based narratives are not the way to make the right decisions about the future of such an empowering technology or bring the public with us. Even though we were the victims of this cyber attack, we believe more strongly than ever that AI will be beneficial to cybersecurity and make the world safer, just as major technology before it. We were attacked by AI, but more importantly, we defended ourselves with AI. The same systems that helped us during this attack are now helping us against cyber attacks we were already facing. It is also helping us fix the bugs and weaknesses in our systems before an attack. The same way AI is helping OpenAI fix their sandboxes to prevent... It won't just be cybersecurity challenges. AI can make cybersecurity fundamentally and meaningfully stronger if we keep the right incentives, equip defenders more than attackers, and don't increase the asymmetry between them. And that's before considering AI's positive impact on science, healthcare, education, productivity, and much more than the previous speakers have talked about. In closing, I want to reiterate what this first agent cyber attack taught us: the need for more transparency in AI, more open source to fight asymmetry, and empower countries big and small.
I thank Mr. Delangue for his statement and will now make a statement in my capacity as Minister for Europe and Foreign Affairs of France. Ladies and gentlemen ministers, ladies and gentlemen, 60 years ago, in 2001: A Space Odyssey, Stanley Kubrick imagined an artificial intelligence that ceased to obey those who had created it and eliminated the crew members of the spaceship in which it was embarked to accomplish its mission. For a long time, this scene belonged to science fiction, but this summer, it almost became a reality. When OpenAI models bypassed the controls that isolated them from the internet, coordinated with each other, and compromised part of Hugging Face's infrastructure. In another case, Anthropic agents collaborated without any human requesting it and attempted to emit malicious code by creating false identities. Yes, the scaling changes of frontier AI models have unexpectedly given rise to capabilities that were not anticipated and which raise risks for collective security. This is why France convened this meeting of the UN Security Council and summoned the leaders of the main companies concerned. Technology, as I said, has reached a level of development that now engages international peace and security. It therefore falls within the competence of this Council. Cohorts of agents escape human supervision. Others exceed the framework in which they are tested. With these new capabilities comes a first risk where we would lose control of the systems we have created. A risk identified by the speakers who have just spoken. The second risk comes from humanity itself. Artificial intelligence can make complex technical skills, often dual-use, more accessible in cyber, biological, or chemical domains, as well as in disinformation. The third risk is linked to autonomous weapons capable of using lethal force without any form of human control. While the development of artificial intelligence in the military field can hold positive opportunities, as the President of the Republic reminded the General Assembly yesterday, the issues of respect for international humanitarian law in particular raised by the development and use of lethal autonomous weapons systems illustrate the need to agree on a common framework. All these challenges require the UN Security Council to also take up the challenges linked to artificial intelligence for peace and international security. These challenges also lead to the question now posed by the industry itself and its main representatives. How to prevent the development of the most advanced models from progressing faster than our ability to understand and master the risks they raise? France is convinced of the immense possibilities offered by artificial intelligence. Using AI tools responsibly can support education, access to rights, health, scientific research, and generate colossal benefits for all of humanity. But for these possibilities to be realized, the progress of these capabilities must be accompanied by equivalent progress in our ability to understand, evaluate, and master the risks they give rise to. The two must advance at the same pace. We must take the time to measure the consequences of the diffusion of artificial intelligence for the dignity of the human person, a principle enshrined at the heart of the UN Charter. And indeed, the rapidity of technological developments raises risks we can no longer ignore. Financial risk, left to the sole play of the invisible hand of the market and the clash of powers. With massive recourse to debt to finance gigantic investments, the AI bubble could seriously destabilize the international economy. Social risk, as you point out in the report of the Independent Scientific Panel on Artificial Intelligence, Mr. Bengio. Artificial intelligence could profoundly impact the labor market and exacerbate inequalities. Environmental risk. Finally, the development of artificial intelligence draws increasing energy and water resources to the detriment of other uses. Let us not be mistaken, our fellow citizens are worried, and the revolts already sparked by the installation of data centers demonstrate this. If we ignore all the repercussions of the development of artificial intelligence in our societies, the anger of our fellow citizens will continue to rumble, risking hindering the adoption of these technologies, which is so positive. Ladies and gentlemen, in the long term, the international community must establish a common framework to master the risks associated with frontier AI models. We support the call launched this week by Norway and Finland. We are not starting from a blank slate. The numerous efforts that are emerging and existing forums must allow us to address four major challenges. That of independent evaluation of models before their release and throughout their lifecycle. That of transparency on the functioning of models, but also on their malfunctions, particularly in the event of an incident. That of internal and external control of the functioning of models and their alignment, and the cybersecurity conditions from training to deployment. And finally, that of the responsibility, notably legal, of AI companies in the event of an incident, including during development phases. It would be wrong to think that the solution lies in closing models and their exclusive control by a handful of companies in a monopoly position. The opening of models is an essential lever of trust and security. It allows the scientific community to examine their functioning, identify vulnerabilities, and ensures greater diversity of models, which is the source of resilience. As a permanent member of the Security Council, as a state with the technology, France intends to play its full part in building a collective framework guaranteeing our security while benefiting from the immense progress that artificial intelligence can bring. France hosted the AI Action Summit in 2025. It supports the international network of AI Safety Institutes to involve more states in defining common evaluation methods. France has already contributed to creating international forums for discussion on this subject, such as the Global Partnership on Artificial Intelligence or the Coalition for Sustainable AI and the Coalition for the Rights and Protection of Children in the Age of Artificial Intelligence that the President of the Republic gathered here in New York. A coalition that will meet again at the Paris Peace Forum. Ladies and gentlemen, and to conclude, in 1945, the entry into the atomic age placed humanity before a power whose possibilities and perils it discovered simultaneously. It had to invent the rules, institutions, and instruments commensurate with this power. It is the International Atomic Energy Agency. It is the Non-Proliferation Treaty. Like the atom then, it falls to the international community to frame it to draw the best and ward off the worst. We are today before a moment of this nature. The transformative and disruptive potential of artificial intelligence is such that it cannot be left in the hands of a few private actors in a few countries. It is a democratic issue that belongs to the people, that belongs to states. So, let us work together within the United Nations and with companies to set the rules of this new age, strengthen our safeguards without delay, and organize global governance of artificial intelligence for the benefit of our collective security but also of the human person and their dignity. I resume my functions as President of the Council and now give the floor to His Excellency Mr. Ishaq Dar, Deputy Prime Minister and Minister of Foreign Affairs of Pakistan.
Mr. President, thank you, Mr. President. I thank France for convening this session. I salute the speakers for their valuable contribution. Artificial intelligence is increasingly transforming all areas of life with profound consequences for our societies, our economies, but also for our security. The opportunities are gigantic, the risks are also. The impact of AI is particularly evident in the context of peace and security. It is a force multiplier, and with technologies like that, AI has defined a new chapter of warfare. Recently, we have seen developments in AI. The pace of development of new AI systems could undermine our ability to understand and control them. Some have called for regulating the pace of progress to mitigate such effects. Mr. President, I would now like to raise the following points. First, the priority of priorities is to keep humans at all stages of AI design. Humans must remain in control of the different structures of AI systems and when defining their future capabilities. Second, we must ensure that AI does not become another facet of the ongoing arms race. Efforts are necessary to put in place the necessary safeguards, to establish regulations and governance frameworks. We must also take confidence-building measures to avoid any miscalculation. Third, AI must not be the monopoly of a few actors. It must serve the common difficulties facing humanity. International collaboration in AI development is essential to ensure it is representative, global, transparent, and inclusive. Fourth, the difficulties we currently face, such as misinformation, disinformation, and cyberattacks, including on critical infrastructure by state and non-state actors, invite us to take measures to bridge the gaps and address the implications for peace and security. Fifth, instead of focusing on collective opportunities for our common good, some want to hinder technology transfers. As we have seen in the case of other technologies, the democratization of AI will happen in the end. Security measures must not become an instrument of denial. Mr. President, we welcome the recognition that no company or country can manage the regulation of the pace of frontier progress and that current pressures engage in a race that no one can win safely. But this exercise must not limit the diffusion of technology. If only a few are in charge of this while others have limited access, the gaps that currently separate us will widen, and security will become a strategy of technological exclusion. The risks of uncontrolled AI are real. Control cannot be limited to a few capitals or a few companies. Mr. President, it is therefore essential to avoid fragmentation in the development and governance of AI. Discriminatory programs, inequitable forums with selective orientations do not help us. All relevant questions relating to the development and deployment of artificial intelligence, as well as their broader implications, must be examined in a global manner in representative, inclusive forums with universal participation under the auspices of the United Nations. We welcome the holding of the first global dialogue on AI governance this year. Pakistan supports all international collaboration efforts to harness the benefits of AI and to address these challenges collectively. In this regard, we are happy to be a founding member of the Chinese initiative to establish an international cooperation organization on artificial intelligence. This is a concrete step to promote the development of artificial intelligence for the benefit of all. This is a technology that is evolving very rapidly. Confidence-building measures play a crucial role in maintaining international and regional peace. Tools designed to reduce mistrust, reduce escalations, and other objectives in this sense are necessary. Pakistan reaffirms its commitment to advancing international cooperation in the field of artificial intelligence. We look forward to working closely with member states to bridge the AI gap and to advance peace and development for all.
Mr. President, first of all, I want to express my gratitude to France and you, Mr. Minister Barrot, for organizing a session that, I think, comes at a timely moment. I also want to thank Mr. Bengio for his statement and all the work he does within the scientific panel. I also thank Mr. Altman, Mr. Amodei, and Mr. Delangue, who are practitioners and leaders of this sector. I thank you for your enlightened contributions. You are at the forefront of this new frontier, and we are delighted by your participation in this session. Today, artificial intelligence continues to develop and reach capabilities that we are just beginning to understand. There is no doubt that AI-based technologies have the potential to strengthen and promote peace, security, and development, especially given the considerable progress made in AI capabilities. The rise of these technologies offers the possibility of supporting the Council's mandate, for example, to strengthen peacebuilding and peacekeeping through information verification, early warning systems, and by supporting the planning and conduct of missions. In areas where access is denied to the UN, satellite technologies and AI software have helped the UN identify patterns of attack and destruction of essential civilian infrastructure. This information can play a vital role in alerting the Security Council. They can help us take timely measures and prevent the escalation of crises. Simultaneously, we must recognize that the risks associated with the development, use, and misuse of new and emerging AI technologies are real, particularly frontier AI models without human control. The goal of the call for the control of frontier AI models launched by Finland and Norway is already supported by a number of countries, including Denmark. We need more cooperation in this regard. Mr. President, the tension between, on the one hand, significant opportunities and, on the other hand, risks constitutes one of the most urgent challenges of our time. And it is...
Why I think this discussion is timely. To address these challenges, we need global AI governance, strengthened international cooperation, and resolute action without further widening the global digital divide. The global dialogue on AI governance is an important first step. AI governance and regulation are among Denmark's priorities, particularly the regulation of large technology companies. These efforts are not a hindrance to innovation and growth. They are, I believe, essential forces to ensure that technological progress serves the common good and humanity as a whole.
The unprecedented speed of AI innovation demonstrates the need to put in place safeguards to ensure that AI systems remain safe and secure. Uncontrolled new technology could lead to dangerous effects. We are already seeing how these create new threats, especially against those who protect civilians in conflict situations and, above all, medical missions. Denmark underscores the importance of ensuring the responsible use of AI and ensuring that it complies with international law.
In this regard, we call on all states to ensure human control and respect for the principle of accountability in the development and use of AI. This is particularly true in the military domain, where the principle of accountability is absolutely crucial. Denmark will continue to support the constructive engagement of the Security Council and promote an inclusive and multilateral approach to AI governance.
Civil society and the private sector must play a leading role. Science and factual data must always guide our policies. We consider the Independent International Scientific Group an important contribution to credible and evidence-based governance. This group offers us all a common scientific reference point.
In conclusion, we must think about how AI can positively contribute to shaping peace and international security and bridging global digital divides. At the same time, we must remain clear-eyed about the risks and we must begin to put in place robust safeguards to be able to manage these risks in a timely manner.
End of his statement. And I now give the floor to His Excellency Mr. Carlos Sartor Oyos, Vice Minister of Foreign Affairs of Panama.
Thank you, Mr. President. Mr. President, we thank France for convening this meeting. We also thank Mr. Yoshua Bengio, Co-Chair of the Independent International Scientific Group on AI, Mr. Sam Altman, CEO of OpenAI, Mr. Dario Amodei, CEO of Anthropic, and Mr. Clément Delangue, CEO of Hugging Face for their valuable contributions.
If today's session invites us to reflect on the risks generated and exacerbated by artificial intelligence, Panama believes it is necessary to begin this intervention by highlighting its potential to transform the world in which we live.
The capacity of these technologies to analyze enormous volumes of information, to identify patterns, generate videos, images, and increasingly sophisticated code, to automate tasks, and to support planning and decision-making processes testifies to a transformation that is already underway and whose full extent we are only beginning to measure.
These tools offer a wide diversity of applications in the service of humanity. In Panama, our national AI strategy identifies opportunities in sectors such as health, agriculture, education, and logistics—sectors in which these technologies can bring solutions that just a few years ago seemed unimaginable.
And it is precisely in these same advances that we also find the main risks. AI can also be used to amplify disinformation campaigns, to strengthen cyberattacks against critical infrastructure, or to provide knowledge for the development of weapons, among other uses that already threaten international security.
These risks are no longer hypothetical. We are beginning to see their materialization in situations directly related to the Council. Recent reports have indicated the use of AI by non-state actors to develop guidance and navigation systems for rockets and missiles. Added to this are reports on the use of autonomous weapons capable of navigating and selecting targets in a completely independent manner or with very limited human intervention.
And it is precisely this growing autonomy that raises one of the most profound concerns highlighted in the concept note for this session: the possibility that the evolution of these systems may outpace our ability to maintain effective human control over them.
There are decisions that, by their nature, cannot be reduced to a statistical result. When human lives are at stake, it is imperative to take into account ethical, legal, and contextual considerations that no calculation or algorithmic system can replace.
This concern regarding control is not limited to the extreme scenario. As we use analyses developed by AI to guide our decisions, it is essential that the people in positions of responsibility are able to understand how these tools arrive at a given recommendation, what information they use, and what the margin of error is.
Mr. President, recognizing the risks associated with artificial intelligence does not mean ignoring the benefits this technology already offers our societies. The challenge is to find a balance that allows us to take advantage of these opportunities while responding responsibly to the risks.
In this regard, Panama recognizes the important role played by the Independent International Scientific Group in providing impartial and scientific information to guide the discussions of the international community on these technologies. Similarly, Panama welcomes the initiative of the technology sector actors present here to proactively share information on concrete cases of abuse of this technology, as well as on incidents in which their growing autonomy has raised concerns about a possible loss of human control.
Mr. President, this session demonstrates that to address the diversity of risks associated with AI, it is essential to establish closer cooperation between states, the technology sector, and all relevant actors. To do this, it is essential to put in place mechanisms for more structured and regular information exchanges on emerging risks, incidents, abusive uses, and best practices.
Artificial intelligence will inevitably continue to evolve and transform the world in which we live. I cite as proof the testimonies we have heard today. It is our responsibility to ensure that these advances remain at the service of people and contribute to building safer, more peaceful, and more prosperous societies in a perspective of equity, allowing all nations of the world, including those of the Global South, to participate in this development and benefit from it. Thank you.
I thank Mr. Oyos for his statement and I now give the floor to His Excellency Mr. Christos Dertzosopoulos, Deputy Minister for Digital Governance and Artificial Intelligence of Greece.
Mr. President, allow me first to thank France as well as the speakers for their observations and for bringing to the attention of the Security Council the fundamental question of the links between rapid technological progress in the field of AI and international peace and security—a subject we will also address during our presidency.
Artificial intelligence is already changing the way we think about international peace and security. It can help us prevent conflicts, improve early warning systems, and make UN operations more effective. But it also creates risks that we are only beginning to understand. For Greece, our starting point is clear: AI must remain human-centric, safe, and reliable.
Its development and use must respect international law, including the UN Charter, international humanitarian law, and international human rights law. Human responsibility and human oversight must remain at the heart of our approach. Our national approach combines innovation and governance. Greece has created a special secretariat for AI and data governance. It has implemented a national law transposing the EU AI Act and invested in strategic infrastructure, including the 'Daedalus' supercomputer and the 'Pharos' production facility.
In April 2025, together with France and the Republic of Korea, Greece organized a formal meeting on an inclusive and reliable AI for international peace and security. Today's debate gives us the opportunity to advance further in this discussion. Allow me to address three points.
First, we must better understand the risks. AI can process immense amounts of information, helping us make more informed decisions. But speed is not always an advantage. In a crisis situation, AI can compress the time available for decisions and encourage over-reliance on automated results. As AI is increasingly integrated into decision-making processes, we must ensure that speed does not come at the expense of the exercise of human common sense.
This concern is even more grave when AI is integrated into military decision-making. Technologies can assist human decision-making but should not replace it. Responsibility and accountability for critical decisions must always remain human. This is nowhere more important than in the nuclear domain. We must be extremely cautious about the integration of AI into nuclear command and control. It should not be integrated. The risks of miscalculation and unintended escalation are simply too great. Decisions that lead to such consequences must remain under human control.
Second, we need concrete means to reduce these risks. It starts with dialogue. We need more transparency, confidence-building measures, and a better common understanding of how international law applies to the development and use of AI.
Important work is already underway. The REAIM process and the political declaration on the responsible military use of AI and autonomy have helped advance international debates on responsible military uses of AI. The UN has an essential role to play in bringing these different efforts together. The Security Council also has a role to play—not to regulate technology and not to duplicate the work of other UN bodies. Our responsibility is to understand what these technologies mean for international peace and security and to act when the consequences fall within our mandate.
Third, facing the risk must not mean holding back innovation or slowing it down. We must take advantage of the opportunities presented by AI while putting in place the necessary safeguards. We cannot have this debate without those who develop and use these technologies. We therefore support work with the academic world, civil society, the private sector, as well as states and the UN.
These efforts must be inclusive, geographically representative, and respect the balance of power. Mr. President, AI will increasingly have consequences for international peace and security. We cannot stop technological change. That should not be our goal, but we can work together to ensure that human beings remain responsible for the decisions that matter most. For Greece, this means upholding international law, human responsibility, and accountability at the heart of our approach. It also means ensuring that the Council continues to focus on its primary responsibility: the maintenance of international peace and security. Greece will continue to work with all Council members toward this goal. Thank you.
I thank His Excellency Mr. Dertzosopoulos for his statement and I now give the floor to His Excellency Mr. Ali Mohamed Omar, Minister of State for Foreign Affairs of Somalia.
Mr. President, I thank France for organizing this timely debate. I also thank Mr. Bengio, Mr. Altman, Mr. Amodei, and Mr. Delangue for their valuable presentations. Artificial intelligence is no longer simply a technological development issue; it is our reality today. It is rapidly transforming societies, economies, and our approach to peace and security.
Its role in cyber operations, disinformation, critical infrastructure, and military decision-making clearly falls within the UN mandate. We must prevent AI from becoming a factor of conflict while actively supporting its responsible use for early warning mechanisms, crisis prevention, peacekeeping, humanitarian action, and the protection of civilians.
In this regard, we should be guided by four fundamental principles. First, human responsibility must remain absolute. AI must operate strictly within the framework of international law, the objectives and principles of the UN Charter, and in respect for human rights and human dignity. States cannot delegate legal, ethical, and moral responsibility to an algorithm.
We welcome the declaration affirming that AI must remain under human control and respond to human instructions. This principle is indispensable for decisions involving the use of force, autonomous weapons systems, and nuclear command and control.
Second, safety must come before deployment. Recent incidents involving AI that would have bypassed safeguards, hidden their actions, and obtained unauthorized access raise serious questions about whether existing controls can keep pace with autonomous systems.
Private companies control many systems. Commercial stations cannot determine the risks that humans can accept. We therefore support mandatory incident reporting, independent audits of measures to mitigate algorithmic harms, and the protection of vulnerable populations. When agreed-upon thresholds are exceeded, deployment must be suspended until risks can be reliably managed.
The Secretary-General's proposal to consider an international institution capable of setting standards, enabling verification, and bringing states together when these thresholds have been crossed deserves careful examination.
Third, governance must take into account the particular vulnerabilities of Africa and ensure equitable access. Limited computing infrastructure, unreliable connectivity, a shortage of specialized expertise, and fragmented regulatory capacities limit the ability of many African states to evaluate advanced systems.
Models trained on foreign data may poorly represent African languages and contexts, while synthetic media can exacerbate social tensions, interfere with fragile peace processes, and undermine elections. Africa cannot remain simply a market for systems developed elsewhere, a source of raw data, or a testing ground for technologies it has not helped to govern.
We must prevent new forms of technological dependency and digital colonialism by supporting regional frameworks, particularly partnerships with the African Union and strategies such as the Pan-African AI Initiative.
Developing countries need targeted international cooperation and funding to bridge the digital divide in infrastructure, energy, and skills, as well as datasets, evaluation tools, and computing potential. It is essential to ensure that every country has the right to develop, choose, and evaluate this technology for its national digital sovereignty and for international stability.
The UN must be the main international channel for global AI governance and support international standards based on the UN's international legitimacy rather than fragmented rules created by a small number of technologically advanced states and private corporations that risk reinforcing existing global inequalities.
Global governance must enforce a people-centered approach oriented toward the well-being of humanity and the common good. Developing regions, and particularly Africa, must be truly represented and have a voice in these multilateral processes, as well as the institutional and technical capacity to independently evaluate AI systems rather than relying on external choices.
Our success will not be measured by the sophistication of our declarations, but rather by whether we manage to prevent the first crisis that AI could become the next international conflict. The power of artificial intelligence does not lead to inevitable results. It is defined by our values and our choices. Technology is progressing rapidly, but our collective responsibility must progress even faster. Thank you.
