While Western tech circles debate whether AI should slow down, China's state security establishment has issued its own risk checklist. Chen Yixin, Minister of State Security and secretary of its Party committee, published a signed article in the latest issue of China Cyberspace magazine, "Comprehensively Building an AI Safety Barrier for Orderly Development," systematically laying out six security risks AI poses to China.
Chen wrote that AI has become the main battlefield of global tech competition and a new track of major-power strategic rivalry — China must seize opportunities while recognizing risks and holding the bottom line. The six risks he listed: first, systemic impact on the political security environment, as hostile forces abuse AI to produce harmful information and wage public-opinion and cognitive warfare; second, disruptive escalation of cyber offense-defense, with AI sharply lowering the barrier to cyberattacks and threatening critical information infrastructure; third, large-scale data-leak and espionage risks, both from foreign spying and domestic misuse of foreign AI tools; fourth, monopoly-driven tech imbalance, as some countries impose entity lists and technology controls under the guise of "national security," severing global AI supply chains; fifth, structural shocks to social governance, where algorithm "black boxes" and data "poisoning" amplify social bias and break traditional governance frameworks; sixth, fundamental change in the form of military warfare.
The article arrived at a pivotal moment in US-China AI competition. It stresses adhering to "Party leadership over the internet and data," accelerating construction of an AI security risk prevention and control system, improving regulatory platforms, intensifying core-technology development such as computing chips, and strengthening the rule-of-law framework. According to The New York Times, this is among Beijing's most systematic, highest-level articulations of AI security risk — while in the United States, the debate centers on whether AI could slip human control; in China, official discourse emphasizes AI's threat to political stability, ideology and critical infrastructure.
This difference in perspective is itself part of the AI governance contest. When Washington calls for industry self-regulation and slower development, Beijing stresses "development and security in parallel" and institutional regulation; when the US accuses China of "industrial-scale distillation," Beijing counters that Washington's technology controls are severing global chains. The two AI superpowers' different definitions of "risk" guarantee their governance paths will diverge.
Chen's article offers a clear coordinate for reading Chinese AI policy: in the official narrative, the core of AI security is not "technology out of control" but "governance under control." This means any US-China dialogue on AI must first bridge their fundamentally different understanding of what "risk" means.