Analysis by CNN's John Liu
(CNN) -- Last weekend, as the world worried about the possibility that unchecked artificial intelligence could wipe out humanity, the leaders of the United States and China laid out their visions for this era-defining technology.
They came from almost opposite sides.
Ignoring calls from AI scientists and executives from Anthropic, OpenAI and Google to slow development, US President Donald Trump said what was more important was to maintain the US lead against China in the escalating great power race in artificial intelligence.
The US President said on Saturday: “We are superior to China in the field of artificial intelligence... Frankly, I want the situation to remain that way, because whoever wins in the field of artificial intelligence will win.”
Chinese leader Xi Jinping, on Sunday, in turn called for strengthening international cooperation. In a speech he delivered in India before leaders of the “Global South” countries, he said that countries should “encourage open sources, openness, cooperation and knowledge sharing,” at the same time reiterating his proposal to conclude a global agreement to govern artificial intelligence, although he did not provide details about what this framework might look like.
The contrast between Trump and Xi's messages highlights the lack of consensus and deep trust gap between the two superpowers in a high-stakes race to dominate the field of artificial intelligence.
It appears that the United States is increasingly weaponizing its dominance over this sector, as it views it as a strategic asset that should be protected and exported to allied countries, while China - which is lagging behind in this race - is proposing a more open model.
Artificial intelligence will be strongly present at the highly anticipated summit between Trump and Xi in Washington in two weeks. In their meeting held in Beijing in May, the two parties discussed establishing regulatory controls for artificial intelligence, but without achieving significant tangible results. The Chinese Ministry of Commerce announced last week that the two leaders agreed to launch a joint government dialogue on this file. However, given the broader geopolitical competition between the two countries, experts' expectations for any substantive breakthrough remain low.
How serious is the threat posed by China?
Chinese AI has been lagging behind its American competitors for years. This is partly due to US technology controls that limit Chinese companies' access to advanced AI processors, which are essential for training increasingly complex AI models.
But last year, DeepSeek, a Chinese AI start-up, surprised the global industry with a model whose performance was close to that of leading American systems, using, it claimed, far fewer computing resources.
Since then, the pace of development of Chinese AI labs has accelerated, with Alibaba, Z.AI, Minimax, and most recently Moonshot unveiling similar models capable of challenging the dominance of leading American companies such as Anthropic and OpenAI.
However, in parallel with the developments in Chinese models, American accusations of “cheating” have escalated. Washington—as well as OpenAI and Anthropic—alleged that Chinese companies stole American intellectual property by training their models on the output of leading American models, a process known in the industry as “distillation.”
Last week, US federal agencies, including the Federal Bureau of Investigation (FBI), the National Security Agency, and the Cybersecurity and Infrastructure Security Agency, alleged that Chinese AI companies were stealing “on an industrial scale” from their American counterparts.
In a lengthy report released last week, Anthropic also detailed attempts by “malicious actors” — including those allegedly linked to the Chinese government — to misuse its systems to develop weapons and conduct surveillance on minorities like the Uyghurs in the Xinjiang region.
The report cited an example in which Anthropic disclosed sensitive information related to Chinese surveillance attempts; Queries from a user - believed to be connected to the Chinese military - originally directed to Moonshot's Kimi model were redirected to Cloude's model, revealing a secret attempt by Moonshot to divert user requests to Cloud. The report added that other Chinese companies, such as DeepSeek, also practiced this type of action.
Beijing has rejected these claims, while Chinese AI labs have mostly remained silent. Moonshot declined to comment on the report that reached CNN last week, while Deep Sec did not respond to a request for comment.
Foreign Ministry spokesman Guo Jiaqun said on Monday: “Promoting various threatening narratives, confrontation and malicious competition will only hinder efforts to govern artificial intelligence globally, and will not serve the interests of any party.”
He added: "All parties should work together to promote artificial intelligence that is open, inclusive, and accessible, and benefits humanity."
What is China proposing?
Just as the United States increasingly feels threatened by rapidly evolving Chinese models, Beijing is also concerned about the cyber capabilities of systems developed by American companies; These capabilities were clearly demonstrated this summer when systems from OpenAI and Anthropic went out of control and launched unauthorized hacking attacks against other systems.
Beijing also views AI safety risks from a different perspective, going beyond the American concept of existential risks, such as AI systems getting out of control. In addition to prioritizing political control and maintaining social stability, Beijing sees the United States as a threat that hinders its own development of this technology.
In an article published on Sunday, the Minister of State Security, Chen Yixin, highlighted the dangers of “hegemonic practices,” a veiled reference to the United States and criticism of its actions of “establishing lists of entities, imposing technological restrictions, monopolizing industry standards, and creating closed source systems.” This is under the pretext of protecting national security.
“These actions fragment the global AI industry and supply chains, exacerbating the development imbalance within this industry globally,” the Chinese minister wrote.
An editorial published over the weekend in the Global Times, a popular state-backed newspaper, downplayed a recent article by Dario Amodei, CEO of Anthropic.
“At first glance, this appears to be a ‘rational statement’ focused on global AI security, but a closer reading reveals its hidden agenda. The article is filled with containment provisions targeting China and is essentially a Cold War scenario applied to the AI sector,” the editorial said.
Are people in China concerned about artificial intelligence?
While prominent executives and researchers in the United States have openly expressed concerns about the risks of this technology, raising similar concerns in China, where the ruling Communist Party strictly censors the media and surveils the population, is much more difficult and carries far greater risks.
China's technology sector also operates within a much more "top-down" regulatory environment, with the government setting comprehensive rules governing what AI systems can produce and how they can be deployed. For example, China's national safety standards for artificial intelligence set requirements that include training data, privacy, security, and prohibited content.
There is also greater general optimism about AI in China; A 2025 Edelman survey showed that 72% of respondents in China said they trusted AI, compared to just 32% in the United States.
Although Chinese artificial intelligence laboratories have quickly closed the gap with their American counterparts, they still lag behind the most advanced American systems. This may reduce the incentive for Chinese companies and researchers to voluntarily restrict the development and deployment of increasingly capable AI systems as they play catch-up.
Can the two powers agree on artificial intelligence?
There seem to be few reasons for optimism; Despite the trade truce reached between the two countries in October last year, tensions continued to escalate.
The main obstacle is trust; Washington fears that cooperation will give China access to technology that enhances its military and economic power, while Beijing sees “national security” as an American pretext for curbing China’s technological rise.
A post published last month on Yuyuantantian - a social media account affiliated with the official Chinese Broadcasting Corporation (CCTV) - set the terms of the US-Chinese dialogue on artificial intelligence, indicating that Washington must ensure that its companies working in this field adhere to any rules and regulations that are agreed upon, as a precondition for any negotiations.
“A clear distinction must be made between real threats to safety and mere technological competition,” the statement said. “The fundamental question is not ‘should regulation be imposed’, but rather who should set the safety limits for these pioneering cutting-edge models.”
