The United States and China are competing for leadership in artificial intelligence, but the two powers are also confronting risks that neither country can easily manage alone. As U.S. President Donald Trump and Chinese leader Xi Jinping prepare to meet, AI safety is emerging as one possible area for dialogue despite continuing disagreements over advanced chips, national security and technological competition.
Rival Powers Face a Common AI Problem
Competition between Washington and Beijing has intensified as China narrows parts of the technological gap with the United States. Both governments see artificial intelligence as strategically important, while also worrying that increasingly capable AI systems could create new security threats.
Scott Singer, co-director of the China AI Initiative at the Carnegie Endowment for International Peace, said the AI relationship between the countries has become closely tied to broader tensions between Washington and Beijing.
Those tensions make cooperation difficult, even when the two governments have concerns about many of the same potential dangers. These include AI-enabled cyberattacks, biological misuse, failures involving advanced models and situations in which humans could lose control of powerful AI systems.
Building an Emergency Line for AI
One possible area of cooperation involves establishing ways for the United States and China to communicate during serious AI-related incidents.
U.S. Treasury Secretary Scott Bessent said American and Chinese officials discussed creating an AI dialogue mechanism during talks in New York. The United States also proposed a notification system that could be used when an AI incident reaches the level of a national security concern.
“We want a shared vision of common goals and common threats,” Bessent told reporters.
Such communication could become particularly important if AI is used to attack critical infrastructure. Chinese AI governance documents have identified areas including finance, electricity, telecommunications and transportation as potentially vulnerable to risks involving advanced AI.
Experts also say China has demonstrated interest in discussing AI and cyber incident information-sharing with American specialists. Regular communication during emergencies could gradually create greater technical confidence between the two countries.
Competition Over Chips Remains a Major Divide
While safety offers some possibility for cooperation, the technological race remains a major obstacle.
Trump has emphasized maintaining the United States’ advantage in AI. Washington has also imposed restrictions designed to limit China’s access to some of the world’s most advanced computing technology and chips.
Beijing has criticized those restrictions and accused the United States of attempting to maintain an AI monopoly. Analysts cited by AP expect significant relaxation of advanced-technology export controls to remain difficult as both governments pursue strategic advantages.
Despite the restrictions, Chinese AI companies have continued advancing their models. Systems developed by companies including DeepSeek and Moonshot AI have competed with leading American models on some measures while sometimes operating at lower costs. Chinese technology companies are also increasingly developing and using domestically produced chips.
Professor Will Cong of Nanyang Technological University said export restrictions have not prevented China from developing competitive frontier models, although they have constrained the computing resources available for expanding their use.
AI Technology Becomes a Security Dispute
Another source of tension involves allegations surrounding “distillation,” a technique through which one AI model can learn from the outputs of another.
The U.S. government and American AI companies have raised concerns that Chinese developers have extracted capabilities from leading American systems. Chinese officials have rejected accusations of malicious activity, while some analysts argue that forms of distillation are widely used throughout the AI industry.
Anthropic has accused Chinese AI developers including DeepSeek and Moonshot of routing some user requests through its Claude models. The dispute has added another layer to an already complicated relationship in which commercial competition, intellectual property and national security increasingly overlap.
China has its own concerns about American AI systems, including the possibility that powerful models could threaten critical infrastructure or contribute to political misinformation.
Safety Cooperation Meets Strategic Distrust
The central challenge is that AI safety discussions are taking place at the same time as both governments increasingly view advanced technology through a national security lens.
China has shown growing concern about risks including malicious AI use, cyberattacks, autonomous agents and the possibility of losing control over advanced systems. American officials have similarly highlighted risks that could affect both countries.
This creates an unusual situation: Washington and Beijing have strong incentives to outperform each other technologically while also having reasons to prevent catastrophic AI incidents that could harm both countries.
As Trump and Xi meet, major disagreements over chips, AI models and technological leadership are unlikely to disappear. However, establishing reliable communication during serious AI emergencies could provide a limited area where the two governments’ interests overlap. The broader question is whether competition for AI leadership will leave enough room for cooperation when the risks cross national borders.
