Washington and Beijing opened a government-to-government channel on advanced artificial intelligence risks, but they have not published any rules for how it will work.
Story Snapshot
- U.S. and Chinese leaders endorsed talks on artificial intelligence risks and safety.
- Officials held their first bilateral artificial intelligence meeting in Geneva in May 2024.
- Leaders discussed military artificial intelligence uses and responsible development.
- No public agreement defines how either side must notify the other about incidents.
Leaders Back Artificial Intelligence Risk Talks
White House records say President Joe Biden and President Xi Jinping agreed to address artificial intelligence risks through direct talks. The April 2, 2024 call readout said they would have experts discuss artificial intelligence risk and safety. A November 16, 2024 meeting readout said they affirmed the need to improve artificial intelligence safety and international cooperation. These documents show top-level backing, which often decides whether lower-level channels get traction.
These leader statements matter because process shapes outcomes. When presidents set a direction, agencies tend to follow. In a tense rivalry, even basic agreement to talk can be rare. The commitment does not settle disputes over trade, technology, or security. It does create room to design steps that can reduce accidents. That includes sharing definitions, testing hotlines, and setting thresholds for alerts. But none of that exists until both sides write it down and try it.
First Meeting Exchanged Views, Not Rules
Officials from both countries met in Geneva on May 14, 2024, for the first artificial intelligence dialogue. The United States described the talks as candid and constructive. China called them in-depth and professional. The Associated Press reported that the meeting grew from the leaders’ prior summit and covered each side’s approach to safety and risk management. U.S. officials also raised concerns about alleged Chinese misuse, while China criticized U.S. restrictions.
That exchange shows the narrow path for cooperation. Each side wants guardrails that cut the risk of a sudden crisis. Each also wants to protect its own programs and its advantage. Without a written protocol, the channel remains a venue to talk, not to act. There is no shared rule on who must notify whom, when, or about what. That gap makes it hard to judge whether the talks can stop a fast-moving problem before it grows.
Military Artificial Intelligence Is On The Table
The White House November 16, 2024 readout says the leaders discussed developing artificial intelligence in the military field in a prudent and responsible way. Analysts have linked these talks to worries about how software could affect command decisions for critical weapons. Some track this concern to risks like autonomous weapons or systems that could misread data and push humans toward bad calls. The official text does not spell out nuclear command links, leaving details unclear.
Still, the military angle raises the stakes. When machines speed up sensing and targeting, small errors can spread fast. History shows rivals build hotlines and rules to slow that spiral. The United States and the Soviet Union used crisis lines during past wars to reduce miscalculation. Washington and Beijing later set up their own links after earlier tensions. Artificial intelligence adds new pressure to make those links work under stress.
What A Working Channel Would Need
A usable channel needs clear terms. Both sides would have to define what counts as an artificial intelligence incident, who sends alerts, how fast they must send them, and what facts they must share. They would also need a way to test the system in drills. The public record so far does not show a signed protocol with those steps. The effort competes with export controls, sanctions, and public blame that make sharing harder.
For Americans across the spectrum, the stakes are simple. If a software glitch or spoofed signal triggers a false alarm, leaders need a trusted way to say, “Stand down, here are the facts.” That protects troops, trade, and families at home. It also limits chances for elites to hide behind vague promises. The talks set a floor, not a ceiling. The test now is whether Washington and Beijing will publish rules, drill them, and prove the line works when it counts.
Sources:
warontherocks.com, thediplomat.com, breakingdefense.com, cc.pacforum.org
