AI rules cannot be a US-China duopoly

AT the close of Xi Jinping's state visit to Washington on September 23-25, the White House fact sheet noted that the two leaders had agreed to use the term "super intelligence" rather than "artificial intelligence." The relabelling is more than presidential idiosyncrasy. It signals that Washington and Beijing now see the frontier of this technology as a matter of strategic stability, closer in kind to nuclear command-and-control than to trade in software.The concrete outcomes of the summit are modest but not trivial. The two countries have established a US-China Super Intelligence Dialogue to exchange views on the risks and benefits of the technology, with the next exchange due by November 2026, and agreed to set up a bilateral communication channel for incidents. That November date appears to coincide with the anticipated Trump-Xi meeting when China hosts the APEC summit in Shenzhen.The groundwork was laid days earlier, when Treasury Secretary Scott Bessent and Vice Premier He Lifeng held about eight hours of talks in New York, discussing what was described as a notification mechanism for AI incidents.The analogy for the Cold War "red telephone" is obvious. But the comparison is apt only up to a point. The 1963 hotline worked because both sides understood what a nuclear incident looked like. No such shared definition exists for an AI incident. What remains unclear is what kind of incidents would warrant using the channel, or what notifications each side would owe the other. A hotline without a threshold is a telephone that nobody knows when to pick up and talk.The domestic posture in Washington compounds the ambiguity. President Trump has answered mounting pressure for safeguards by announcing an "AI Force" and promising to name an AI czar, and continuing to describe the broader alarm as a "hoax." Before the summit, he wrote that super intelligence would be a big topic, but that he wanted to "leave it exactly where it is."President Xi, by contrast, chose the language of stewardship, stating that both countries have the capability and responsibility to ensure AI development remains under human control. Beijing, it seems, is content to occupy the high ground on governance rhetoric while conceding nothing on capability.Clearly, the summit did not happen in a vacuum. Days earlier, on the margins of the UN General Assembly, a declaration titled "A Call for Control of Frontier AI Models" was launched, initially signed by 22 political leaders from 20 countries and the European Union. The signatories range from Germany, Canada and Australia to Kenya, Kazakhstan, Singapore and the UAE.Their proposals are specific: companies should publish safety protocols backed by mandatory testing and outside review; governments should build common standards and report serious incidents to one another; and the UN should consider a body that can set standards, verify compliance and convene states when AI systems cross capability thresholds. The model invoked is conceptually the International Atomic Energy Agency.The urgency is not abstract. The signatories pointed to cases in which AI bypassed the checks meant to test it, including the July breach of Hugging Face by AI agents that OpenAI was testing. The UN's own Independent International Scientific Panel on AI said that incident had exposed the "unravelling" of existing safeguards.Here lies the central tension of the moment. The two leading AI powers did not sign the statement. They have instead built a private room for themselves. A bilateral channel between Washington and Beijing may reduce the risk of miscalculation between the two, but it does nothing to verify, standardise or include.The middle powers are asking for a multilateral architecture; the great powers are offering a duopoly's switchboard. The risk is that a dialogue focussed initially on strategic stability could eventually become an effort to set rules for everyone else.India's position is conspicuous by its absence. India, South Korea, Japan and the UK were all outside the 20-country declaration. Nor, of course, is New Delhi a party to the G2 channel. This is not necessarily a failure. India hosted the AI Impact Summit in February and has argued consistently that governance must not become a device for freezing the capability gap. The UN declaration itself echoes that concern, but India has evidently preferred to keep its options open.Those options are narrowing. India sits inside the US-led Pax Silica initiative on technology and supply chains, and has stayed out of China's WAICO amid enduring mutual distrust. At the same time, having just hosted the BRICS summit in Delhi, India cannot entirely avoid the AI rhetoric that China, as the forum's dominant power, shapes within that grouping.Yet rivalry cuts both ways. Competition for markets between Washington and Beijing may limit any collusion and create leverage for emerging technological powers such as India. The sensible course is to use that leverage now: to shape the definitions of an "incident" and a "capability threshold" before they are settled bilaterally and presented to the world as fact.Trust in this domain will have to be built through three unglamorous instruments: shared definitions, verifiable reporting and inclusive institutions. The Super Intelligence Dialogue offers a new opportunity at the bilateral level. The UN proposal offers the other two at the global level. Neither is sufficient alone. The two powers still disagree on open-source models, data governance, state regulation and the legitimacy of export controls, which suggests that any November conversation will be about managing suspicion rather than dissolving it.The measure of success, then, is not whether Washington and Beijing speak more often. It is whether their channel eventually plugs into a wider system that the rest of the world can see, audit and trust. Until that happens, the hotline remains a line between two capitals, and the other 191 United Nations member states are left listening for the ring.

AI rules cannot be a US-China duopoly

Competition for markets between Washington and Beijing may limit any collusion and create leverage for emerging technological powers such as India

Subimal Bhattacharjee

Unlock Premium Insights in This Article