AI security debate enters national security narrative, Chen warns of "hostile forces" using artificial intelligence

China's Foreign Ministry on September 14 publicly rejected Anthropic chief executive Dario Amodei's demand for continued restriction of advanced AI chips and equipment exports to China; at the same time, China's National Security Minister, Chen Yin, published an article that advanced AI could be used by "hostile forces" for cyber attacks, intelligence collection and ideological security, directly into the national security framework.

The two groups of statements appeared in advance of the Qinghai meeting on September 24.The AP that AI governance is expected to be one of the topics of Trump and Xi's talks, but the Chinese-U.S. definition of "security" is clearly different: U.S. technology companies and security agencies emphasize cutting-edge models of out-of-control, cyber attacks and technology spread risks, while Beijing is increasingly linking AI security with technology blockade, intelligence activities and political system security.

A technology security debate is being redefined as national competition

Amodei said the world should slow the pace of development of cutting-edge AI capabilities and continue to restrict the flow of state-of-the-art AI chips and manufacturing equipment to China. he believes that if China takes the lead in cutting-edge AI, it could bring geopolitical and security risks, while acknowledging that global security rules ultimately require China’s involvement.

China’s Foreign Ministry spokesman Guo Jianquan responded that “threat, confrontation and vicious competition” targeted by China would only undermine global AI governance, and called on all parties to cooperate.

The argument exposes an unavoidable contradiction: on the one hand, the United States wants China to participate in the cutting-edge AI security mechanism, on the other hand, it restricts China’s access to state-of-the-art computing power; on the other hand, Beijing initiates global AI cooperation, and on the other hand puts foreign advanced models and national security threats into the same set of political narratives.

Chen directed AI risks directly to “political, institutional and ideological security”

Chinese Ministry of National Security Minister Chen Yin Jin subsequently issued an article, saying that artificial intelligence if held in the hands of "hostile forces and other people" could directly threaten China's "political security, institutional security and ideological security". he also mentioned that advanced models can be used to detect vulnerabilities on a large scale, perform complex cyber attacks and collect sensitive data.

It means AI risks in China’s national security system include not only infrastructure attacks, business secrets and personal data breaches, but also challenges to political order and ideological control.

When “model security” and “institutional security” are placed within the same national security framework, the object of AI regulation is no longer just code and computing power, but may also extend to information dissemination, model output and social political expression.

During Xi Jinping's tenure, China's national security concept continuously expanded to the fields of economy, science and technology, data, education, culture and public opinion.After AI was incorporated into this system, regulatory policy is likely to simultaneously bear multiple goals such as industrial development, content management, data control and political security.

U.S. security agencies rank Chinese model developers as risk of technology theft

The other side of the controversy is the recent accusations of U.S. security agencies against Chinese AI developers.The Associated Press that the U.S. Federal Bureau of Investigation, the National Security Agency and the Cybersecurity and Infrastructure Security Bureau had previously issued a joint alert, accusing some Chinese developers of extracting capabilities from U.S. advanced models through "damping" and describing related activities as aggressive and malicious.

The Chinese Ministry of Commerce denied the allegations, calling "distillation" a technology method widely used in the industry, and accused the United States of attempting to monopolize the AI industry. Technical model distillation itself is indeed a common type of method, disputed about specific data, how to access and whether it violates terms of service, intellectual property or cybersecurity laws.

Both sides linked each other’s industrial policies with national security policies: the United States worried that advanced computing and modeling capabilities entered China’s military, intelligence and cyber attack system; while Beijing believed that U.S. chip control and security rules were meant to block China’s technological upgrading.

Tensions Between “Open AI” and Political Control

In recent years, Xi has advocated the establishment of a global AI governance framework and depicted China as a driver of open source and universal AI. The low-cost model launched by Chinese enterprises has also expanded its international influence.But domestically, generated AI services are still bound by systems such as content regulation, data security and algorithmic records.

Chen’s latest statement further shows that for Beijing, AI is not just an industrial competition, but also a national regime security issue.This security logic could drive more model review, data localization, protection of critical infrastructure, and restrictions on cross-border exchange of technology.

When the meeting is about to discuss AI, the two countries are not faced with a simple "whether or not to regulate", but with a security system of mistrust of each other.The United States is concerned that China will form a strategic advantage after acquiring cutting edge capabilities; while China believes that the United States will use security rules to consolidate its technological advantage and see external AI capabilities as tools that may affect political and ideological security.

If the two countries do not first establish a minimum level of risk definition and verification mechanism, so-called global AI security cooperation could easily become another technology blocking front to compete with national security.