Beijing’s spy chief focuses on the risk posed by AI to Party rule, not to humanity.

By Probe International

When China’s minister of state security writes that artificial intelligence could pose a direct threat to the Chinese Communist Party’s hold on power, the statement should be read as policy, rather than rhetoric.

Chen Yixin is not a commentator. He runs the Ministry of State Security, the Party’s principal instrument for domestic political security and foreign intelligence.

A recent article by Mr. Chen, published by the state-run journal China Cyberspace, does not dwell on distant catastrophe. It names the assets to be protected—political, institutional, and ideological security—and the instruments of harm: deepfakes, foreign models, data leaving the country, attacks on infrastructure, military disadvantage. It then prescribes a familiar remedy: more Party leadership over the industry, and stricter oversight. [See: China’s Top Spy Chief Warns A.I. Is a Threat to Party Rule]

Mr. Chen’s article is the fullest public warning yet from the Ministry of State Security. He said people with “ulterior motives” could use AI-generated video and audio to wage a “propaganda war,” including deepfakes of officials—even a false Xi announcement—that spread faster than the state can contain them. He wrote that AI had slashed the cost of cyberattacks and posed “serious risks” to China’s information infrastructure, a point underscored when researchers disclosed a tool that could seize millions of WeChat accounts. He warned that Chinese users of foreign models could leak sensitive data; Anthropic said the Chinese start-up Moonshot AI had routed user queries to Claude that included military-linked surveillance and proprietary corporate information. Mr. Chen also argued that warfare had been “fundamentally transformed,” citing U.S.-Israeli use of AI against Iran, and added that whoever masters algorithms for sensing and targeting “will gain the initiative on the battlefield.”

Mr. Chen’s article appears as Washington and Beijing prepare AI talks and a Trump–Xi meeting, and speaks to what is a non-negotiable for China: data sovereignty and political control.

Against a backdrop of mounting concern, American lab chiefs such as Dario Amodei warn that AI could escape human control, help build biological weapons, and even outpace their utility as a tool—treating part of the risk as common to both superpowers. In its coverage, the New York Times contrasted U.S. loss-of-control fears with Mr. Chen’s warning about Party rule.

For the latter, the question is whether AI stays under human control. In China, the real threat is whether the public stays under Party control.