The United Nations Independent International Scientific Panel on Artificial Intelligence, a 40-member body established by the General Assembly in August 2025 and co-chaired by Turing Award recipient Yoshua Bengio of the Université de Montréal and Nobel laureate Maria Ressa, released its preliminary report on Wednesday, calling itself the first global independent scientific assessment of A.I. and warning that the window for coordinated governance “remains open but may not stay that way for long.” The launch lands five days before diplomats convene in Geneva on July 6–7 for the inaugural U.N. Global Dialogue on A.I. Governance, an event whose political ceiling was set the day before publication, when the Trump administration used a U.N. Security Council debate to reject exactly the kind of oversight the panel is designed to inform.
The report’s central diagnostic is structural rather than moral. More than 40 A.I. governance frameworks already exist worldwide, most safety evaluations are performed by developers on their own models, and most countries, including advanced economies, lack the technical capacity to independently evaluate frontier systems or participate meaningfully in their regulation. Meanwhile the compute base sits in two capitals: roughly 75 percent of the world’s leading A.I. supercomputing power in the United States, about 15 percent in China, nearly 90 percent between them.
That distribution is the entire game, and every actor at Geneva knows it.
Michael Kratsios, director of the Office of Science and Technology Policy, explicitly rejected “centralized control and global governance” of A.I. at the Security Council session, extending a posture that saw Washington decline to sign the Paris summit declaration earlier this year. The break from precedent is sharp: the same administration, in its first term, endorsed the OECD A.I. Principles in 2019 and helped found the Global Partnership on A.I. in 2020. Beijing has read the opening and taken it, backing a U.N.-anchored framework and warning against governance becoming “a game of the club of wealthy nations.” Iraq, speaking for the G77 and China, pushed for the Geneva agenda to address infrastructure and data-access disparities directly.
The panel is scheduled to deliver a fuller assessment in 2027, with members serving three-year terms. Tony Oweke, a researcher at the Council on Foreign Relations, has framed the near-term stakes plainly: if the Geneva dialogue stalls, governance defaults to “exclusive multilateral clubs.” The vocabulary of legitimacy is already being contested, and the country holding three-quarters of the compute has decided it prefers the contest to the consensus.
Sources
- https://www.usnews.com/news/top-news/articles/2026-07-01/un-report-sees-enormous-potential-benefits-and-big-risks-from-ai
- https://news.un.org/en/story/2026/07/1167848
- https://www.un.org/independent-international-scientific-panel-ai/en/preliminary-report
- https://www.cfr.org/articles/the-world-is-trying-to-govern-ai-the-un-wants-in
- https://www.csis.org/analysis/what-un-global-dialogue-ai-governance-reveals-about-global-power-shifts