On February 12, the United Nations General Assembly approved a 40-member scientific panel to study the global impacts of artificial intelligence. The vote was 117 to 2. The two countries that voted against it were the United States and Paraguay.
Russia voted yes. China voted yes. Every EU member voted yes. Japan, South Korea, India, Brazil, the UK, Australia — all yes. The only abstentions came from Ukraine, which objected to a Russian-nominated expert, and Tunisia.
The panel is modeled after the IPCC, the Intergovernmental Panel on Climate Change that spent three decades building the scientific consensus that eventually produced the Paris Agreement. The AI version will have three years to do the same thing for artificial intelligence: issue annual evidence-based reports on what the technology is doing to economies, labor markets, information ecosystems, and weapons systems. Its 40 members were selected from over 2,600 candidates by the ITU, UNESCO, and the UN Office for Digital and Emerging Technologies. They include two Americans — Vipin Kumar and Martha Palmer — and Filipino Nobel laureate Maria Ressa.
The United States didn't just vote no. It called for the recorded vote in the first place.
Lauren Lovelace, a counselor at the U.S. Mission, called the panel "a significant overreach of the U.N.'s mandate and competence." She said AI governance "is not a matter for the U.N. to dictate" and warned against "ceding authority over AI to international bodies that may be influenced by authoritarian regimes seeking to impose their vision of controlled surveillance societies."
At the India AI Impact Summit a week later, Michael Kratsios, the White House's top technology adviser, was more direct: "AI adoption cannot lead to a brighter future if it is subject to bureaucracies and centralized control."
Sam Altman, standing at the same summit, said the opposite: "Centralization of this technology, in one company or country, could lead to ruin. We obviously do need regulation or safeguards, urgently."
Secretary-General António Guterres framed the panel as the antidote to both extremes. "Less hype, less fear," he said. "More facts and evidence." He called for making "human control a technical reality — not a slogan."
The pattern here is familiar. When the IPCC was established in 1988, the United States participated. It helped write the reports. It shaped the framework. Three decades later, the Paris Agreement reflected American scientific contributions even when American politics didn't. The IPCC's authority didn't come from enforcement power. It came from being the place where the evidence lived.
The AI panel is designed to work the same way. No binding regulations. No enforcement mechanisms. Just 40 scientists producing annual reports that every country — including the ones building the technology — will have to respond to publicly.
That's precisely why the U.S. voted against it.
The seven companies that control frontier AI development are all American. OpenAI, Anthropic, Google DeepMind, Meta, xAI, Microsoft, and Amazon are collectively spending over $300 billion on AI infrastructure in 2026 alone. An independent scientific body publishing annual assessments of what that spending produces — and who it affects — creates a baseline that lobbyists can't rewrite.
The U.S. position isn't incoherent. It's strategic. America leads in AI. International scientific consensus is what turns leadership into liability. When the IPCC published its reports on fossil fuels, the American energy industry didn't dispute the science for long — it disputed the panel's authority to produce it. The objections sound identical: overreach, mandate creep, authoritarian influence.
What makes the AI vote different from climate is speed. The IPCC had decades to build consensus before the politics caught up. AI doesn't have decades. The models that exist today are already reshaping labor markets, information systems, and weapons programs. By the time the AI panel publishes its first annual report, the technology it's assessing will have changed.
The U.S. argued the UN should focus on "core missions" like international peace and security. But 117 countries just said that AI governance is international peace and security. The question the panel will answer isn't whether AI needs oversight. It's whether the country that builds the most AI gets to decide the answer alone.
Two countries think yes. A hundred and seventeen think no.\n\n---\n\n*If you work with AI tools daily, check out my AI prompt engineering packs — battle-tested prompts for developers, writers, and builders.*
Top comments (0)