In a moment that may mark a turning point in humanity's relationship with its own creations, the leaders of two of the world's most powerful artificial intelligence companies appeared before the United Nations Security Council to ask for something unusual: to be governed. Sam Altman of OpenAI and Dario Amodei of Anthropic argued that advanced AI has grown beyond the capacity of any single nation to oversee, and that without international frameworks, the systems being built today could outpace the institutions meant to hold them accountable. Their appeal — made in the chamber where wars and cri
AI Leaders Urge U.N. to Establish Global Framework to Keep Systems Under Human Control
The conversation has moved into formal diplomacy
Mark
Why would Altman and Amodei voluntarily ask for international restrictions on their own work? That seems counterintuitive.
Mimi
Because they believe the alternative is worse. If every country sets its own rules independently, you end up with a fragmented system where companies have to navigate dozens of incompatible requirements. A single global framework, even if restrictive, is actually simpler to operate within.
Luke
But we should be careful here. The source doesn't detail what specific restrictions they proposed or what verification mechanisms they actually suggested. They outlined "the shape" of frameworks, but did they commit to anything concrete?
Mimi
Fair point. They presented principles, not a binding agreement. The Security Council asked about verification and they essentially said "it's solvable if there's political will." That's not a technical answer.
Mark
What's the real barrier to getting agreement among the permanent members?
Mimi
Geopolitical competition. AI is central to military capability and economic dominance. China and the U.S. have opposite regulatory philosophies. Russia hasn't shown interest in constraining itself. Getting five countries with conflicting interests to agree on binding rules is extraordinarily difficult.
Luke
And the source doesn't tell us whether any council members actually committed to pursuing this, or if they just listened politely. We know they asked questions about verification, but we don't know if the hearing moved anyone toward action.
Mark
So this is more symbolic than substantive at this point?
Mimi
It's a threshold moment. The conversation has moved from tech conferences into formal diplomacy. That matters. But you're right—whether it produces actual treaties or regulatory frameworks is still completely open.
Luke
One more thing: the source says Altman and Amodei acknowledged their own development would be constrained by such frameworks. But it doesn't say whether they're willing to accept constraints unilaterally, or only if everyone agrees. That's a crucial distinction.
Mimi
Exactly. They're advocating for rules, but only if they apply globally. If the U.S. agrees to restrictions and China doesn't, that's a competitive disadvantage they probably won't accept.
The Pulse
- Two of the most powerful figures in AI walked into the UN Security Council and asked the world's governments to constrain them — a striking reversal from an industry that long resisted outside oversight.
- The urgency behind their message: AI breakthroughs in one country can reshape global capabilities within months, and no single government or institution currently has the tools to track or control that pace.
- Council members pressed hard on the central problem — verification — asking how the world would actually know if a nation or company violated any agreed limits, a question the executives could not fully answer.
- The geopolitical terrain is treacherous: the US and China hold fundamentally opposed visions for AI regulation, and Russia has shown little appetite for constraining its own development, leaving the diplomatic path forward deeply uncertain.
- Despite the obstacles, the hearing itself signals a shift — AI governance has moved from think tanks and boardrooms into the formal machinery of international diplomacy, and the conversation at the highest levels has now begun.
In a moment that may mark a turning point in humanity's relationship with its own creations, the leaders of two of the world's most powerful artificial intelligence companies appeared before the United Nations Security Council to ask for something unusual: to be governed. Sam Altman of OpenAI and Dario Amodei of Anthropic argued that advanced AI has grown beyond the capacity of any single nation to oversee, and that without international frameworks, the systems being built today could outpace the institutions meant to hold them accountable. Their appeal — made in the chamber where wars and crises are weighed — suggests that AI has crossed a threshold from commercial ambition into the domain of global security.
In a rare and symbolically charged appearance, Sam Altman of OpenAI and Dario Amodei of Anthropic addressed the United Nations Security Council with a shared message: the development of advanced AI has become a matter of global security, and no single country or company can manage its risks alone.
The two executives did not arrive with a finished treaty. Instead, they sketched the architecture of what international cooperation might require — shared testing standards before deployment, transparency mechanisms allowing governments to verify what is being built, and agreements on which capabilities should be restricted or monitored worldwide. Notably, they acknowledged that such frameworks would constrain their own companies' timelines and competitive positions, a concession that gave their appeal unusual credibility.
The Security Council's decision to take up the question was itself significant. A body convened for wars, terrorism, and regional crises devoted formal attention to AI governance — a signal that world powers are beginning to classify the technology not as a commercial frontier, but as a matter of international peace. Council members asked pointed questions about enforcement and verification, and the executives offered no clean answers, only the argument that solutions were possible if political will existed.
The appearance also reflected a visible shift within the industry. For years, major AI companies resisted external oversight, favoring self-regulation. That posture has changed. Altman and Amodei's willingness to advocate for binding international rules suggests that some industry leaders now see a fragmented global regulatory landscape as a greater threat to their operations than coordinated standards would be.
The hardest obstacles remain. The five permanent Security Council members hold deeply incompatible views on AI regulation, with the United States and China in particular approaching the technology through competing strategic lenses. Whether the formal machinery of international diplomacy can produce meaningful agreements is far from certain. But the conversation has begun at the highest level — and the people building these systems are now explicitly asking for the constraints that might slow their own progress.
In a rare appearance before the United Nations Security Council, two of the world's most influential artificial intelligence executives made a direct appeal for global coordination on a problem they say cannot be solved by any single nation or company: keeping advanced AI systems answerable to human judgment.
Sam Altman, who leads OpenAI, and Dario Amodei, chief executive of Anthropic, addressed the council with a shared conviction that the trajectory of AI development demands international frameworks—not as a distant possibility, but as an urgent necessity. Their message was straightforward: without coordinated action across borders, the systems being built today could outpace the ability of any government or institution to understand or control them.
The two executives framed their appeal in terms the Security Council understands: global security. They argued that the risks posed by advanced AI systems transcend traditional national interests. A breakthrough in one country's laboratories can reshape capabilities worldwide within months. The economic and military implications are profound. But so, they suggested, is the opportunity to establish guardrails before those capabilities become too complex to govern.
Altman and Amodei did not present a finished proposal. Instead, they outlined the shape of what international cooperation might look like: shared standards for testing AI systems before deployment, mechanisms for transparency that allow governments to verify what companies are building, and agreements about which capabilities should be restricted or monitored globally. They acknowledged that such frameworks would constrain their own companies' development timelines and competitive advantages—a concession that lent weight to their argument that they believed the stakes justified the cost.
The Security Council hearing itself was significant. The body typically convenes to address military conflicts, terrorism, and regional instability. That it devoted time to AI governance signals a shift in how world powers are beginning to categorize the technology: not merely as a commercial innovation or a scientific frontier, but as a matter of international peace and security. Several council members asked pointed questions about verification—how would the world actually know if a country or company was violating such agreements? The executives offered no definitive answers, only the observation that the problem was solvable if the political will existed.
The appearance also reflected a broader repositioning within the AI industry itself. For years, the largest AI companies resisted external oversight, arguing that self-regulation and market competition were sufficient safeguards. That posture has shifted noticeably. Altman and Amodei's willingness to appear before the Security Council and advocate for binding international rules suggests that at least some industry leaders now believe that the alternative—a fragmented global landscape where different countries impose incompatible requirements—poses greater risks to their business models than coordinated international standards would.
What remains unclear is whether the Security Council's five permanent members—the United States, Russia, China, France, and Britain—can actually agree on AI governance principles. The technology is central to military strategy and economic competition. China and the United States have fundamentally different approaches to regulating AI development. Russia has shown little interest in constraining its own capabilities. The diplomatic challenge is immense, and the executives offered no illusions about that difficulty.
Still, the hearing marked a threshold. The question of how humanity governs artificial intelligence has moved from academic conferences and corporate boardrooms into the formal machinery of international diplomacy. Whether that machinery can produce meaningful agreements remains to be seen. But the conversation has begun at the highest levels, and the people building the technology are now explicitly asking for the constraints that might slow their own progress.
Notable Quotes
The executives argued that without coordinated international action, advanced AI systems could outpace the ability of any government or institution to understand or control them.— Sam Altman and Dario Amodei, addressing the UN Security Council
Want the full story?
Read the original at The New York Times ↗
Technology