Skip to main content
Advertisement
Live broadcast
Main slide
Beginning of the article
Озвучить текст
Select important
On
Off

The United States and China have been racing in the field of artificial intelligence (AI) for a long time. "However," previously, everything was reduced to the development of "classic" AI agents, but now the confrontation goes beyond "civilian tasks" — countries are turning this tool into a weapon. Now Beijing and Washington are trying to agree on ways to jointly monitor the activities of AI. What is the essence of the race between the USA and China, how AI tools have transformed over time and what humanity will expect in the future — in the "Izvestia" material.

Robots and humans

US President Donald Trump called the issue of regulating the activities of artificial intelligence one of the central topics discussed with Chinese President Xi Jinping at a meeting that took place on Thursday, September 24.

Трамп и Си
Photo: REUTERS/Evan Vucci

"Big day with Chinese President Xi. Superintelligence will be a big topic of discussion, but I want to leave it as it is. China has the same position," the politician wrote on Truth Social.

On Tuesday, speaking at the UN General Assembly, the American leader also said that the United States would encourage the development of artificial intelligence, not restrain it. This statement was a kind of response to the warnings of AI agent developers that unlimited improvement of models could lead to various global threats.

In turn, Xi Jinping, speaking at the White House as part of a state visit to the United States, noted that it is Beijing and Washington that are responsible for the effective development of AI models, their control and management.

However, if you look at the whole picture, you can see that the United States and China have been competing quite fiercely for the status of the "main AI power" for a long time. More and more analysts have been talking about this lately. The two countries' rivalry generally lies in the control of the so—called "basic technological infrastructure" - chips, data, expert personnel, high standards, markets, etc. The powers compete in everything: in the amount of computing power, the quality of models, industrial and military implementation, and commercialization.

The winner, according to experts, will largely determine which technical protocols, security requirements, principles of data access, modes of openness of models and standards of responsibility will operate in the global digital economy.

Телефон
Photo: Global Look Press/Sergey Elagin

Today, the United States still holds the leading position in the AI race, Artem Egorov, head of the laboratory of immersive educational technologies at the Skolkovo School of Management, told "Izvestia". The American models — Claude Fable from Anthropic and GPT Astra from OpenAI — go far ahead in terms of capabilities and functionality. This is especially evident in agency work, when the AI performs the task itself, and the person connects solely as a supervisor.

— China has taken a different approach: its models are open and cheap. They can be downloaded, deployed on your server and "retrained" for yourself, but the American ones are closed and sold only directly. Fable's million tokens output costs $50, DeepSeek's $1.2 is 40 times cheaper," the expert explained.

He also clarified that Chinese artificial intelligence models were trained specifically on American developments. For example, the DeepSeek R1 AI agent once looked like a copy of the o1 program from OpenAI. However, now, according to the source, the situation has changed: Beijing is rapidly catching up with Washington in terms of development.

— Until the summer of 2026, the gap was large, and then it decreased sharply. In June, GLM-5.2 was released, in July — Kimi K3, and now Chinese developments in agency tasks have equaled or come close to American ones. This forced Western companies to accelerate the production of new models, and such competition, in my opinion, will ultimately work for the development of the entire industry," said Egorov.

Приложение
Photo: Global Look Press/Matthias Balk

At the same time, he added that Europe is still noticeably lagging behind, and in Russia the main players have to deal with difficulties with the purchase of hardware.

And here is the policy

The rivalry between the United States and China in the field of AI is a struggle for global dominance, which has its roots in the period of the formation of American leadership. After World War II, the United States became the world's leading power, forming the Pax Americana international order, in which it acted as the hegemon and guarantor of the system.

At that moment, the Cold War began very successfully for the United States, which, as is commonly believed, ended with the collapse of the USSR. As a result, Washington lost an external "enemy" against whom it could build the very strategy of global leadership. In fact, at that time, the United States remained the only global superpower, which made it necessary to "create" a new competitor, which was China.

The current US-Chinese confrontation, according to Vladimir Vasiliev, chief researcher at the Institute of the USA and Canada, is very similar to the new version of the Soviet-American confrontation, although it differs from it in some aspects. If the Cold War was largely determined by military, strategic and ideological confrontation, then Washington and Beijing's relations are more tied to the economy - the two countries simultaneously compete and remain deeply interdependent.

Деньги
Photo: IZVESTIA/Dmitry Korotaev

— This is where the main paradox of modern American politics arises. The United States views China as a strategic competitor, but at the same time, it has been the American economy that has contributed to its rise for decades. Beijing's integration into world trade accelerated its economic development, while simultaneously creating a huge market, production base, and source of goods and investments for American businesses, the expert explained.

Why so fast

It can be noted that recently, the heads of American technology giants — OpenAI, Anthropic and Google DeepMind — after many years of improving AI programs "suddenly" began to call for a reduction in the pace of its development. Experts are resigning from their positions due to fear of the power of new technologies, and experts estimate the probability of destroying humanity with artificial intelligence at 25%.

On the one hand, industry representatives explain the need to slow down the risks associated with loss of control over technology and insufficient security measures. But on the other hand, Washington may well consider the development of advanced models as a strategic resource, and limiting China's access to advanced technologies as a tool to preserve American leadership.

— Interestingly, according to a Stanford study, America surpasses China in terms of private investment by an order of magnitude, this is $285.9 billion against $12.4 billion in 2025. This is the key difference. In the United States, AI development is provided by private companies, their management is decentralized, the state influences them through procurement, regulation, defense programs and export control, while in China the main customer is the state. This ensures a more extensive and targeted implementation. In addition, there is a noticeable desire in China to reduce dependence on external technologies and strengthen its influence in the Global South," said Ekaterina Movsesian, CEO of CROS.

ИИ
Photo: IZVESTIA/Sergey Lantyukhov

Against this background, the desire of American IT corporations to strengthen state control over technology is beginning to look quite logical. The scheme is quite simple: AI model developers want to maintain their power and influence and at the same time ensure that their projects are protected. In this regard, as the Financial Times points out, American companies are beginning to increase their lobbying activities, seeking to influence the formation of public policy.

At the same time, large-scale investments in unprofitable AI infrastructure are accompanied by an inflating investment bubble. Under these conditions, government support and contracts should help corporations reduce financial risks and maintain access to capital. After all, artificial intelligence in the United States is a private business sector. This creates an interdependence between the state, which is interested in technological leadership and national security, and business, which seeks to maintain profitability and competitive advantages.

The Future of the U.S.-China Rivalry

AI is transforming and, instead of civilian goals, it is beginning to switch to "military tracks." China is already introducing an army of terminator robots to the world, and the United States is announcing the creation of a new special structure, the "artificial intelligence forces," or AI Force.

In addition, AI already poses a threat to the security of government agencies in various countries. One of the high-profile examples of this was the hacking of the Australian Ministry of Health website by OpenAI AI agents. The intrusion into the system occurred back in June. During internal testing, the program penetrated the closed part of the database of the state healthcare system Medicare to find materials for research on medical expenses. However, even in OpenAI, they only found out about the incident in August.

ИИ
Photo: IZVESTIA/Sergey Lantyukhov

Now the United States is trying with all its might to maintain hegemony and turn artificial intelligence into a weapon. Washington is striving to form the Pax Silica technology ecosystem in order to maintain its dominance in the IT sector and weaken its geopolitical competitors. Beijing has taken a different path — it is developing various formats of cooperation with countries, the purpose of which, however, is also to control the development of AI. A striking example of this is the creation of the World Organization for Cooperation in the Field of Artificial Intelligence (WAICO) in July 2026.

It turns out that the price of the US-China race is safety. This is the main factor that does not keep up with the constant improvement of AI agent models. Hence the loud warnings from OpenAI and Anthropic about the risks of using the technology. The threat of uncontrolled AI development is real, but today it is very convenient to use this fear to intimidate all participants in this market. Regulation in the United States, Europe and Australia will first hit Chinese models, which are likely to be "taken away" from users.

— It is difficult to predict what this rivalry will lead to, however, it can be expected that two interconnected but partially separated AI circuits will form. But even with the strategic rivalry between the United States and China, they will have to look for minimum rules of interaction. For example, measures to prevent dangerous cyber incidents, control autonomous systems, and prevent uncontrolled escalation," concluded Ekaterina Movsesian.

Переведено сервисом «Яндекс Переводчик»

Live broadcast