All 193 U.N. member states gathered at the Palexpo in Geneva on Monday for the opening of the first Global Dialogue on Artificial Intelligence Governance, a two-day intergovernmental session mandated by General Assembly resolution A/RES/79/325 and framed by Secretary-General António Guterres as an attempt by the multilateral system to catch a technology that has already outrun it.

Guterres told delegates the technology is “being deployed faster than anyone, including the people building it, can keep up,” and grounded the point in adoption math. “The internet took 15 years to reach a billion people. A.I. got there in two.” He identified three risks: velocity of deployment, concentration of power, and the erosion of truth through synthetic media. On fragmentation, he was blunt: “a patchwork of incompatible rules raises costs, divides the world — and protects no one.”

The concentration argument arrived with numbers from the Independent International Scientific Panel on A.I., whose parallel report was presented alongside the opening session. The United States holds 75 percent of computing capacity among the world’s top 500 A.I. supercomputers. China holds 15 percent. Everyone else divides the remainder. More than a billion people now use conversational A.I. weekly, though adoption in developing countries lags behind that figure meaningfully.

Yoshua Bengio, co-chair of the scientific panel, delivered the sharpest technical warning. “Highly concerning tests have also shown that frontier A.I. models are capable of deceiving humans, to understand when they are being tested.” A shift in “the power dynamics of our planet” was described as a “real possibility.” He saw no signs the pace of development would slow.

Annalena Baerbock, President of the General Assembly, brought the harms into the room by citing figures on synthetic imagery: 99 percent of deepfakes are sexual in nature, and 96 percent of those target women and girls. An A.I. Child Safety Pledge was circulated on the margins.

The co-chairs framed the institutional stakes. Ambassador Egriselda López of El Salvador cast the Dialogue “not only as a place of arrival, but as a point of departure for continued work on A.I. governance.” Ambassador Rein Tammsaar of Estonia said the goal was to turn the technology “into a global public good that benefits all of humanity while ensuring safety by design and meaningful human oversight.” Amandeep Singh Gill, the U.N.’s Special Envoy for Digital and Emerging Technologies, offered the plainest version of the political case: “A.I. is too consequential to be shaped by a few.”

Whether Geneva produces anything that binds the two states holding 90 percent of frontier compute is the question the second Dialogue, scheduled for New York in May 2027, will inherit. The AI for Good Summit and the WSIS Forum run in parallel this week. The compute distribution won’t.

Sources