Both governments say there will be an AI dialogue from November and a channel for AI incidents. Washington calls it super intelligence, Beijing calls it AI, and neither text mentions chips.
· 5 min read · Band Rules and society · The Pulse, 25 September 2026
Chinese President Xi Jinping made a state visit to the United States from 23 to 25 September. Both governments have now published what the two leaders agreed, and both lists include the same thing on AI: a regular dialogue between the two countries on the risks and benefits of advanced AI, with the next exchange in November 2026, and a direct channel between them for reporting AI incidents.
The White House fact sheet, dated 25 September, calls it the U.S.-China Super Intelligence (SI) Dialogue and says the channel is for "SI incidents". It also says the two leaders "agreed to use the term 'super intelligence' rather than 'artificial intelligence'" to describe these technologies. China's foreign ministry published its own list of eight "deliverables and understandings" the next day. Item seven reads: "The two sides agree on the establishment of the China-U.S. AI Dialogue to exchange views on risks and benefits related to AI." It says the next exchange will occur in November 2026 and that the two sides will set up "a bilateral communication channel for AI incidents".
Both governments put the same agreement in writing, and the wording already differs. The US text uses super intelligence throughout. The Chinese English text uses AI throughout, and the Chinese-language account of the leaders' talks uses 人工智能, the ordinary Chinese term for artificial intelligence. Neither text says what super intelligence covers, or whether the dialogue is limited to the most capable AI systems.
The idea came out of earlier talks. Tom's Hardware reported on 22 September, citing Reuters, that after US Treasury Secretary Scott Bessent and Chinese Vice Premier He Lifeng met in New York the weekend before, the two sides proposed that the leaders discuss AI safety measures, including a hotline for AI incidents. It said the aim was to warn each other about rogue AI systems and about autonomous systems whose actions the other side could mistake for an attack. The texts the two governments published do not describe how the channel will work, who runs it on each side, or what counts as an incident.
Xi set out China's position in the talks, according to the Chinese readout: "要保持人类对人工智能技术的管控" (humans must keep control over AI technology). He also said the two countries "应该各展所长而不是相互设防" (should each play to their strengths rather than guard against each other). The same readout says Trump told Xi that AI "关乎人类未来" (concerns the future of humanity) and that the two countries should keep talking about it. Wang Yi, China's foreign minister, said the leaders agreed to "用好中美人工智能对话机制" (make good use of the China-US AI dialogue mechanism), wording that treats the dialogue as something already running rather than new.
What is missing matters too. Neither government's published account mentions chips or controls on selling AI technology to China. The only export controls in the White House fact sheet cover two chemicals used to make fentanyl. The US text does say the two sides "continue to work on U.S. concerns regarding supply chain shortages related to rare earths and other critical minerals".
The November meeting is the first test: whether it takes place on time, who attends for each side, and whether either government publishes what was discussed. The second is the incident channel. It becomes something real once there is a named office on each side and a written definition of an AI incident.
None of the four government documents says who will lead the dialogue, whether the US use of super intelligence is meant to narrow the subject to the most powerful systems or only changes the name, or whether the channel covers military uses of AI. China's list separately says the two militaries will conclude a memorandum on crisis communication and prevention as soon as possible; it does not link that to AI. Reuters' own report could not be read by this desk, so the account of the New York talks relies on Tom's Hardware's summary of it.
For now, this is an agreement to talk, with a date. It sets no rule for any AI company or product, and neither government's published account touches the rules on which AI chips can be sold to China. If the November meeting happens and the channel gets a written procedure, it will be the main line the two governments have agreed to use when an AI system causes trouble that reaches across both countries.
Everything above is written from these. Each line says what that document proves.
· 5 min
The accord is one page of text, voluntary, and asks each company for four layers of checks, one of them an outside auditor. It came on the same day Trump ordered federal agencies to call AI 'Super Intelligence'.
· 4 min
Spain's new AI plan targets 100,000 AI accelerators by 2028 or 2029 in a gigafactory that is still a bid in a European call closing on 12 November.
· 5 min
The order does not build anything. It gives a panel two months to say how a shutdown mechanism for frontier models could be made law, and verified.
Strata, the whole AI stack, explained simply. Every number carries a source and a confidence label.
Today · The Stack · Stories · Archive · Glossary · Break the Chain · Focus · About · RSS feed
Privacy · Terms · Corrections · Report an error
Also from Dheeco: Dheeco · AP Fact Check
Published by dheeco.com · © 2026 Dheeco