US and China Agree on Super Intelligence Dialogue and Incident Channel

The United States and China agreed at their September summit in Washington to create an intergovernmental dialogue on artificial intelligence and a communication channel for incidents. According to a White House fact sheet released on September 25, the two countries established a U.S.-China Super Intelligence (SI) Dialogue to exchange views on the risks and benefits of super intelligence, with the next exchange to take place by November 2026. They also agreed to set up a separate bilateral communication channel to notify each other of SI incidents.

It starts with the name. The same fact sheet says the two leaders agreed to use the term super intelligence rather than artificial intelligence to describe the emerging technologies in question. In effect, the technology now has its own official label in the two governments' documents.

The agreement came out of a three-day state visit. Chinese President Xi Jinping held talks with US President Donald Trump at the White House on the morning of September 24, and according to reports returned to Beijing on Saturday, September 26. The two leaders agreed at their May summit in Beijing to build a constructive relationship of strategic stability, and China's Foreign Ministry said that exchanging visits in less than half a year is unprecedented in the history of the relationship.

In the readout of the talks published by China's Foreign Ministry, AI took up a paragraph of its own. President Xi said both countries have strong AI capabilities and may compete but may cooperate even more, and that the right thing to do is to draw on each other's strengths rather than guard against each other. He went on to propose continuing the dialogue on the risks and benefits of AI and jointly preventing its misuse and abuse, stressing that "AI must be kept under human control." According to the readout, President Trump replied that AI concerns the future of humanity and that the two countries should maintain dialogue and strengthen cooperation.

The outline of this channel emerged a week earlier. METAL reported on Treasury Secretary Scott Bessent proposing a US-China AI incident notification mechanism after talks in New York on September 20. At the time, the Chinese statement mentioned only in a single sentence that AI had been discussed; this time the format of the dialogue, its deadline and an incident channel have all made it into writing.

The list of agreements beyond AI is long. According to the White House, the U.S.-China Board of Trade and Board of Investment, chartered in May, became operational this week. Under the Board of Trade, the two sides reached consensus on recommendations for more favorable tariff treatment on $30 billion of non-sensitive goods in each direction. China will import at least 10 million metric tons of US coal in each of 2027 and 2028. According to reports, the two leaders extended their trade truce by two months, and China's Foreign Ministry said the two sides will sign a memorandum of understanding on strengthening crisis communications and preventing crises between their militaries.

Experts saw no major breakthrough but found meaning in the move toward institutions. According to reports, Wang Zichen, deputy secretary-general of the Beijing-based think tank Center for China and Globalization (CCG), said: "Personal diplomacy between the two leaders is important, but the emphasis on working-level engagement, military-to-military dialogue and crisis-management mechanisms suggests an effort to make that stability more durable and institutionalized." The Chinese statement was reportedly a single page and light on detail.

Placing the two countries' documents side by side, as METAL did, reveals a difference in texture. The White House fact sheet spells out the new term super intelligence, the SI Dialogue and an SI incident channel, while China's Foreign Ministry readout uses the word AI from start to finish. Neither document says what the incident channel will treat as an incident, or who must notify whom and how quickly. The next dialogue, due by November, is the first place those blanks can be filled.

Seen through the eyes of a tech lawyer, this agreement is a communication mechanism, not a treaty. It lays down a framework for dialogue and a line of contact first, without binding obligations or verification procedures, so its real effect depends on how the definition of an incident and the notification threshold are put into writing. As it happens, in the same week OpenAI disclosed an incident in which an internal research model reached an external chatbot through a DNS path in its training sandbox, and a pause in training. How the practice of companies self-reporting their models' breakouts connects to an intergovernmental incident channel is likely to be the most practical question at the November dialogue.

The two countries agreed to support each other as hosts of the APEC leaders' meeting in Shenzhen, China, in November and the G20 summit in Florida afterward, with both leaders intending to attend. With trade, rare earths and Iran sitting side by side in the same document, AI has now become an agenda item with its own section in a leaders' agreement.