Guterres' final UNGA: AI curbs as 15-member Security Council takes up AI

For lawyers and RegTech teams, the signal isn't in what Guterres signed — nothing was signed — but in the institutional machinery now rotating toward AI. A 15-member Security Council preparing to question AI leaders, paired with a US-proposed safety notification mechanism, is the early anatomy of a disclosure-and-testing regime that will eventually reach due-diligence checklists, liability disputes, and cross-border compliance programs. The question is no longer whether AI governance becomes legally binding, but in which forum and on what timetable.

United Nations Secretary-General António Guterres used his final address to the 81st General Assembly on Tuesday, September 22, 2026, to issue a sweeping closing mandate: regulate artificial intelligence, end ongoing wars, make polluters pay, and overhaul institutions he warned are no longer able to confront mounting global challenges. Delivered before nearly 130 assembled leaders in New York — even as Chinese President Xi Jinping and leaders from South Africa, India, and Germany stayed away — the speech positioned AI governance as the clearest test of whether the UN can still shape the rules of emerging technology while major powers increasingly route around it.

“

US Treasury Secretary Scott Bessent said on Sunday, September 20, that Washington had proposed a new AI safety notification mechanism — a disclosure-style framework that would ask developers to flag emerging risks.

The AI component was not rhetorical decoration. Guterres reiterated his call for global oversight of AI risks at the precise moment the UN's 15-member Security Council is preparing to host AI leaders for a briefing on the technology, a step that would drag frontier-model governance out of the advisory arena and into the body responsible for international peace and security. "Those countries leading this technological revolution must share information on emerging safety risks, cooperate on testing and evaluation and work towards common safeguards that protect us all," Guterres told the assembly, a direct appeal aimed at the handful of governments and companies that dominate advanced AI development.

The timing also aligns with a concrete, if still embryonic, American overture. US Treasury Secretary Scott Bessent said on Sunday, September 20, that Washington had proposed a new AI safety notification mechanism — a disclosure-style framework that would ask developers to flag emerging risks. Coming from an administration that has slashed UN funding and skipped much of the multilateral agenda, the proposal signals that even skeptical powers want a hand in shaping AI safety norms rather than ceding the field to the EU's AI Act or Chinese state-led rules. For companies, a notification mechanism of this kind is the thin end of a compliance wedge: voluntary disclosure today, mandatory reporting tomorrow.

The address also laid bare the structural constraints on any binding AI regime. Guterres reflected openly on "gridlock within the organisation," and the empty chairs of Xi Jinping and others illustrated the convening-power deficit at the heart of the multilateral system. A UN that cannot guarantee attendance at its flagship diplomatic meeting will struggle to convert speeches about "curbs" into enforceable obligations. Yet the sustained applause that greeted his defence of a two-state solution, his "polluters must pay" demand, and his institutional-reform push shows there is still an audience — and a political market — for principled multilateralism, even if enforcement levers remain weak.