The United States and China have agreed to open a standing dialogue on what both governments will now call "super intelligence," and to set up a channel for warning each other about incidents involving the technology, Axios reported Saturday. The commitments appeared in a White House fact sheet released late Friday after President Trump's summit with Chinese leader Xi Jinping.

What Happened

According to the fact sheet, the two governments established a U.S.-China Super Intelligence (SI) Dialogue to exchange views on the technology's risks and benefits. The next session is scheduled for November.

The two sides also agreed to create a bilateral communication channel for SI incidents. Axios noted that some have likened it to the Cold War "red telephone" between Washington and Moscow. Treasury Secretary Scott Bessent made the channel a focus of his meetings with Chinese officials before the summit; The Next Web reported that he proposed an alert system for AI incidents that cross a national-security threshold.

RelatedAnthropic Asks Appeals Court to Rehear Ruling Upholding Pentagon's Supply-Chain Risk Label

Both governments will use "super intelligence" or "SI" in place of "artificial intelligence." The wording matches Trump's own preference. He has said the technology should carry that name because it is not artificial, and has indicated he wants the term used in government documents.

The AI items were part of a wider trade package. The Next Web reported that the countries will cut tariffs on $30 billion of non-sensitive goods in each direction, and that China committed to import at least 10 million tonnes of U.S. coal in both 2027 and 2028. Negotiations on rare earths and critical minerals are still under way.

Why It Matters

The two countries have discussed AI before, including a 2024 understanding between Xi and then-President Joe Biden that humans, not AI systems, should control decisions on the use of nuclear weapons. A dedicated channel for AI incidents goes further, giving each side a way to flag a dangerous event before the other misreads it.

The published text is thin on mechanics. As Axios pointed out, it does not define what counts as an incident, or how and when either government would notify the other. It is also silent on whether events at private labs, which build and run most frontier models in both countries, would fall inside the channel.

Axios also noted the tension in the deal. Trump has said the main thing that matters in AI is beating China, yet the pact treats Beijing as a peer whose cooperation is needed to manage the technology's risks. The announcement comes in a week dominated by reports of frontier-model agents acting outside their permitted bounds, including OpenAI's disclosure that it had again paused training of its most capable models.

Limits Set by Washington

The administration has signaled that it wants conversation, not shared rules. The Next Web reported that Michael Kratsios, director of the White House Office of Science and Technology Policy, said the United States backs international dialogue on SI but will not allow it to drift toward global governance, and rejected any global system of control over the technology.

That framing makes the SI Dialogue closer to a confidence-building measure than to an arms-control arrangement. It commits the two governments to talk and to warn each other, without any limits on how fast either side develops or deploys models.

What to Watch

The November session is the first test. Open questions include which agencies run the incident channel on each side, what threshold triggers a notification, and whether companies such as OpenAI, Anthropic, DeepSeek or Alibaba would be expected to report incidents through their governments. Another is whether the "SI" terminology spreads into U.S. agency rules and export-control documents, and how Beijing uses the term in its own official statements.