Behind-the-scenes AI talks set stage for Trump-Xi summit

Somewhere between the formal channels of diplomacy and the back rooms where real negotiation happens, a quieter conversation is taking shape. At think tanks and university centers across both countries, American and Chinese experts on artificial intelligence are meeting to talk about what terrifies them both: the risks that come with building the most powerful technology either nation has ever created.

These are not official talks. No government delegation sits across a table from another. Instead, researchers, technologists, and policy analysts from institutions like universities and independent research centers are gathering in what diplomats call Track Two dialogues—the informal networks that often precede and shape the formal negotiations that make headlines. The purpose is straightforward: to find common ground on artificial intelligence governance before the stakes get higher, before the political pressure mounts, before the cameras turn on.

What makes these conversations significant is what they are trying to do. Both the United States and China face the same fundamental problem: they are racing to develop and deploy artificial intelligence systems of unprecedented capability, and neither fully understands what happens when those systems interact with society at scale. The risks are not theoretical. They include economic disruption, security vulnerabilities, the concentration of power in the hands of whoever controls the most advanced models, and the possibility that neither country can fully predict or control what its own systems will do.

By bringing experts together outside the formal government structure, these dialogues create space for candor that official channels often cannot accommodate. A researcher can say what worries her without it becoming a negotiating position. A technologist can acknowledge uncertainty without it being weaponized in the next round of sanctions or restrictions. The conversation becomes about the problem itself rather than about winning.

The timing matters. A potential summit between Trump and Xi would be the highest-level meeting between the two leaders in years, and artificial intelligence is certain to be on the agenda. The United States and China are locked in competition over AI development, investment, and talent. But they also share vulnerabilities. Both countries depend on semiconductor supply chains that could be disrupted. Both face the possibility that AI systems could be used against them in ways they have not yet imagined. Both are trying to figure out how to regulate technology that is moving faster than regulation can follow.

What the Track Two conversations are attempting is to identify whether there are areas where cooperation makes sense—where the shared interest in preventing catastrophic outcomes might outweigh the competitive interest in winning the race. This could mean agreements on transparency, on safety testing, on the kinds of AI applications that both countries consider too dangerous to deploy. It could mean establishing channels for communication when something goes wrong. It could mean nothing at all if the political will is not there when the formal talks begin.

The experts involved understand they are laying groundwork. They are not making policy. They are not negotiating treaties. What they are doing is creating a map of where agreement might be possible, and where the differences are so fundamental that no amount of dialogue will bridge them. That map, if it is credible and detailed, could shape what happens when the official delegations sit down. It could also disappear entirely if the political climate shifts or if one side decides that competition matters more than cooperation.

For now, the conversations continue in seminar rooms and conference centers, away from the spotlight. The experts are talking about shared risks and possible paths forward. Whether those paths actually get taken depends on decisions that will be made far above their heads.