China and the US are competing for AI dominance but have shared concerns
over safety
[September 21, 2026] By
CHAN HO-HIM
HONG KONG (AP) — U.S. President Donald Trump and Chinese leader Xi
Jinping, leaders of rival nations in a race to dominate artificial
intelligence, are expected to meet this week against a backdrop of
rising concerns about the threats that rogue AI could pose to humanity.
The hopes for finding common ground on concerns around global AI safety
are clouded by competition between the two countries. As China narrows
the technology gap with the U.S., each country has expressed concern
over how AI in the other's hands could hurt them.
“The U.S.-China conversation in AI is getting caught in the ups and
downs in the broader U.S.-China relationship,” said Scott Singer,
co-director of the China AI Initiative at the Carnegie Endowment for
International Peace.
On Sunday, U.S. Treasury Secretary Scott Bessent said after talks in New
York with Chinese Vice Premier He Lifeng ahead of the meeting between
Trump and Xi that the two countries discussed the setting up of a
mechanism on AI dialogue and proposed a new “notification mechanism” for
AI incidents that could impact national security.
Here is a look at some topics that may — or may not — come up for
discussion around AI.
Communicating about AI emergencies is one focus
The agenda could focus on AI risks that could harm both countries, such
as AI-enabled cyberattacks, biological misuse, major model failures or
loss of human control, said Sun Chenghao, a senior fellow at the Center
for International Security and Strategy at Tsinghua University in
Beijing. In that case, “there is still meaningful room for cooperation,”
Sun said.
“It would be a mistake to dismiss the entire safety debate as
geopolitical competition,” Sun said. “China itself is paying growing
attention to loss-of-control risks, malicious use, cyber threats and the
safety of AI agents.”
Bessent said after his meeting with He that the U.S. proposed a
notification mechanism between the United States and China, such as when
there are incidents that rises up to the national security level from
AI. “We want a shared vision of common goals and common threats,” he
told reporters.
Michael Kratsios, Trump’s science and technology adviser, told Fox News
Sunday it’s important to have conversations with China about AI. “The
key thing that we want to talk to them about is avoiding shared risks,
and that’s something that I think we can have a fruitful conversation
on,” Kratsios said.
Both governments are concerned about how AI could be used to attack
critical infrastructure. In China, AI governance documents identify
finance, electricity, telecommunications and transportation as areas
potentially exposed to frontier-AI risks, Sun said.
The Chinese side has shown real interest in discussing issues like AI
and cyber incident information sharing with U.S. experts, said Samm
Sacks, a senior fellow at the Johns Hopkins School of Advanced
International Studies’ Institute for America, China, and the Future of
Global Affairs.
If the two countries can discuss how to report and communicate in the
event of an AI-related emergency, said Xiao Qian, vice dean of the
Institute for AI International Governance at Tsinghua University, “it
can help the two countries to build technical confidence over time.”
US export controls affect China's AI development
Trump has repeatedly said “whoever wins AI, wins.” He has insisted the
U.S. needs to keep a technological edge over China and downplayed calls
by American tech leaders to slow AI's development over safety concerns.
Beijing, in turn, has accused the U.S. of pursuing an AI monopoly as it
curbs China’s AI growth including in restricting its access to the
world’s most advanced chips.
[to top of second column] |
People visit the booth for Chinese AI-computing chip start up
Dongfang Suanxin during the World AI Conference in Shanghai, July
17, 2026. (AP Photo/Ng Han Guan)
 At the meeting, any loosening of
U.S. export controls on advanced technologies related to AI is
likely to be off the table, said Alfredo Montufar-Helu with the
consultancy Ankura. With each country seeking competitive advantages
in so many areas, Montufar-Helu said the room for U.S. and China to
collaborate on AI is “quite narrow."
In his essay this month calling for an AI slowdown, Anthropic CEO
Dario Amodei called on the U.S. to continue seeking ways to limit
Chinese AI development. China's Ministry of Foreign Affairs said
Amodei’s warnings about China were confrontational and a form of
“fearmongering” and that all parties should work together on AI.
Despite the restrictions, Chinese AI models increasingly challenge
American frontier models. Chinese AI models led by Moonshot AI and
DeepSeek this year have shown they can be, by some measures, as
powerful as those of U.S. rivals like Anthropic and OpenAI, often at
a lower cost. Many of them are making inroads globally, including in
the U.S.
“Export controls have not stopped China from developing
frontier-competitive models, but they slow its rate of expansion by
limiting the compute it can use to serve users,” said professor Will
Cong of Nanyang Technological University in Singapore.
Despite the constraints, China's chip developers have been catching
up with the U.S., and companies in China are turning more to
domestic chips. That's giving Xi more confidence in his talks with
Trump, George Chen with The Asia Group said.
Accusations of ‘distillation’ are a sore spot
U.S. claims that China is ripping off American AI technology are
another sore spot in the relationship. The U.S. government and AI
companies have stepped up allegations of “malicious” extraction of
capabilities, or “distillation,” from top American AI models, claims
that Chinese officials have rejected.
Anthropic has accused Chinese AI models of DeepSeek and Moonshot of
routing a number of their user requests to its Claude AI models.
Chinese labs are likely attempting to “piggyback on the massive
training and inference infrastructure of their U.S. rivals” given
how U.S. restrictions are limiting China's compute power, Rebecca
Arcesati, with the Mercator Institute for China Studies, said.
Chinese officials and some technology analysts have argued that
distillation is not uncommon across the AI industry and is not the
main explanation behind Chinese models’ improving capabilities and
ability to innovate.
Beijing, meanwhile, has expressed worries over threats to China’s
critical infrastructure from American AI, including Anthropic’s
powerful Claude Mythos, and more broadly, the risk of AI’s aiding of
spreading of political misinformation and potentially challenging
its governance.
“We cannot expect meaningful cooperation on AI safety if one side
increasingly treats the other side’s technological progress as a
national security threat,” Xiao said. Disagreements between the U.S.
and China “make dialogue even more necessary,” she added.
___
Associated Press journalists E. Eduardo Castillo in Beijing and Josh
Boak in Washington contributed to this report.
All contents © copyright 2026 Associated Press. All rights reserved |
