United States y China They are fiercely competing to dominate one of the technologies that will likely define this century. Chips, data centers, advanced models, robots, and autonomous systems are all part of a race in which neither country wants to fall behind. But something happened this week that reveals another side of that competition.
Donald Trump and Xi Jinping agreed to establish a bilateral communication channel for incidents related to artificial intelligence, according to the agreements announced following the Chinese leader’s state visit to Washington. It will not be just another diplomatic forum. The idea is to establish a mechanism that allows both powers to communicate when an incident involving advanced AI systems occurs that could pose a risk or trigger a crisis between the two countries.
The agreement comes just as the ability of so-called artificial intelligence agents to act autonomously is beginning to cause problems that, until a few years ago, belonged to the realm of science fiction.
A CHANNEL FOR WHEN THINGS GO WRONG
The White House confirmed the creation of a «bilateral communication channel» specifically intended for incidents related to AI.
In addition, Washington and Beijing established a U.S.-China Dialogue on Superintelligence, where both governments will discuss the risks and benefits associated with the development of these systems. The next round is scheduled for November 2026. The comparison with the famous «red phone» of the Cold War is inevitable, even though it is not literally a phone and has not been announced by that exact name.
For decades, the United States and the Soviet Union maintained direct channels of communication designed to prevent an accident, a misinterpretation, or a military crisis from escalating into a catastrophe. Now a different concern has emerged. What happens if an autonomous system takes a dangerous action? What happens if a cyber operation carried out by artificial intelligence appears to originate from a government when in fact it was not authorized? How can we distinguish a deliberate action from an agent’s unexpected behavior?
These are scenarios that have already begun to feature in discussions between Washington and Beijing.
Days before the presidential summit, U.S. and Chinese officials had agreed to work on protocols for AI incidents. U.S. Treasury Secretary Scott Bessent specifically mentioned risks such as uncontrolled agents, non-state actors, and cyberattacks.
AI IS ALREADY A SECURITY ISSUE AMONG SUPERPOWERS
The mere fact that the United States and China consider it necessary to establish this mechanism represents a significant change. The artificial intelligence It is no longer discussed solely as a commercial technology. It is entering the same diplomatic arena traditionally occupied by military communications, strategic weapons, and mechanisms designed to prevent crises among major powers.
The United States and China are both competitors and two of the world’s leading centers for artificial intelligence development. This creates a unique situation: both want to move faster than the other, but neither has any interest in a technological accident triggering an escalation that neither intended to start.
TRUMP NO LONGER WANTS TO CALL IT «ARTIFICIAL INTELLIGENCE»
The agreement also includes a political quirk. The White House reported that both leaders agreed to use the term «Super Intelligence» (SI) in this dialogue. Trump has promoted this term, arguing that the word «artificial» makes this intelligence seem fake. Beyond the name change, the positions of both leaders show significant differences.
Trump has insisted on avoiding restrictions that could slow U.S. development and allow China to gain a technological advantage. Xi Jinping has publicly expressed a more cautious stance. During his visit to Washington, he stated that both countries have a responsibility to develop and manage artificial intelligence for the benefit of society and to ensure that it remains under human control. AFP
The two countries, however, now agree on one fundamental point: they need to be able to communicate quickly when an incident occurs.
WHEN A MACHINE CAN ACT ON ITS OWN
The timing of this initiative is particularly significant. The most advanced systems are no longer simply programs that wait for a question and produce an answer. These new agents can browse the internet, use tools, execute code, interact with other systems, and carry out sequences of actions with increasingly less human intervention.
That autonomy opens up enormous possibilities, but it also introduces a new problem: a machine can perform actions that its own developers did not anticipate. And when these systems operate on global networks, the line between a technological failure and a national security incident can become dangerously blurred. That is precisely the kind of scenario in which a direct channel between Washington and Beijing could prove crucial.
THE NEW BALANCE OF POWER
During the 20th century, the major powers established mechanisms designed to prevent their nuclear arsenals from plunging the planet into an accidental war. The 21st century is forcing them to consider a different kind of risk.
We are still a long way from knowing just how far artificial intelligence will go. Nor is there any evidence that current systems possess «superintelligence» in the sense that science traditionally uses the term to describe intelligence superior to that of humans in virtually all domains. But the United States and China already consider this technology important enough to establish a specific bilateral mechanism to address incidents involving it.
Two countries locked in a race to build increasingly capable machines have just acknowledged that they also need a way to communicate with each other if any of those machines causes a crisis. And perhaps that is the most significant aspect of the agreement. The race for artificial intelligence continues. But the world's two major technological powers are already establishing a line of communication for the day something goes wrong.
