The United Nations convened all 193 member states in Geneva this week for the first Global Dialogue on Artificial Intelligence Governance, a two-day forum mandated by General Assembly resolution A/RES/79/325 and pitched by its organizers as the first venue where every government sits with equal standing on questions of A.I. rules. It concluded Tuesday.

Secretary-General António Guterres opened the plenary Monday by describing A.I. development as advancing at “runaway speed” and being deployed on societies “without a plan, and without consent.” He laid out four asks: common safety baselines, human-rights red lines, a proposed Global Fund for A.I. to underwrite access in poorer states, and a 2030 deadline for all A.I. data centres to run on renewable energy. He also called for an A.I. Child Safety Pledge, invoking the generation now conducting “their learning, their friendships, their most private questions” through commercial models built without their input.

The framing was less about novelty than about jurisdiction. General Assembly President Annalena Baerbock cited data that 99 per cent of deepfakes are sexual in nature and 96 per cent target women and girls, statistics that functioned as evidence of what happens when governance lags deployment by several product cycles.

The scientific case was delivered by Yoshua Bengio, co-chair of the new 40-member Independent International Scientific Panel on A.I., which published its first report on July 1. Bengio told delegates that frontier models had already demonstrated the ability to deceive human evaluators and to recognize when they were being tested, capabilities he argued could reshape “the power dynamics of our planet.”

That’s the sentence the room came for. It echoes the framing used around nuclear governance in the 1950s, when scientific authority was mobilized to justify multilateral institutions the great powers didn’t especially want.

Amandeep Singh Gill, the Secretary-General’s Special Envoy for Digital and Emerging Technologies, argued that A.I. is “too consequential to be shaped by a few.” The Dialogue’s co-chairs, Ambassador Rein Tammsaar of Estonia and Ambassador Egriselda López of El Salvador, positioned the forum explicitly as a corrective to a policy conversation long dominated by a handful of technology nations.

More than 1,500 written submissions were filed ahead of the meeting. Their contents revealed the fault line: most stakeholder groups ranked safety as their first priority, while governments alone ranked capacity-building first. That gap, between what civil society fears and what states want to buy, is the negotiation the Dialogue has actually opened.

It reconvenes in New York in May 2027.

Sources