China’s top intelligence agency just told the world what keeps it up at night, and the answer is artificial intelligence. The Ministry of State Security published a detailed warning on September 13 that AI technologies could be weaponized to threaten the country’s political stability, marking the spy agency’s most comprehensive public statement on the topic to date.
The article, authored by MSS Minister Chen Yixin and published in China Cyberspace, frames AI as a new arena for great-power competition. It’s a notable escalation from previous MSS statements on the subject, which touched on deepfakes in December 2025 and November 2023 but never went this far in spelling out the perceived dangers.
What the MSS is actually worried about
Chen’s concerns fall into three broad categories: disinformation, cyberattacks, and technological dependence.
On the disinformation front, Chen warned that “hostile forces” could leverage generative AI tools, including deepfakes and AI-generated content, to manipulate public opinion and challenge China’s political integrity.
The cyberattack angle is arguably more concrete. Chen argued that AI’s proliferation has lowered the barriers to launching sophisticated cyber operations. Advanced models can help malicious actors discover system vulnerabilities and develop malware more efficiently than human hackers working alone. In a particularly pointed move, the article specifically named US-developed AI models, including Anthropic’s Claude Mythos and OpenAI’s GPT-5.5-Cyber, as technologies that could enable foreign intelligence operations targeting China’s critical infrastructure.
The third pillar of concern centers on technological sovereignty. Chen emphasized that China must maintain independent control over core technologies, with advanced chips singled out as especially critical. Data leaks from AI systems also featured prominently in the warning, reflecting anxiety about sensitive information flowing through models built and hosted by foreign companies.
No brakes, just guardrails
The MSS chief made no suggestion that China should slow down its own AI development. The message was not “AI is dangerous, let’s pump the brakes.” It was closer to “AI is dangerous, so we need to control the technology ourselves and shape the global rules around it.”
Chen’s article did not reference any concerns about AI posing extinction-level threats to humanity. The MSS is worried about AI as a tool in the hands of adversaries, not as a rogue superintelligence. The threat model is espionage and subversion.
Implications across the Pacific
The timing of the warning is notable. It arrives as both Washington and Beijing are grappling with how to regulate AI technologies, and the fact that China’s spy chief and American tech leaders are simultaneously calling for greater caution — albeit for very different reasons — underscores a convergence of anxiety around AI.
For US AI companies, being named in an MSS warning carries real consequences. Anthropic and OpenAI are now officially flagged by China’s intelligence services as potential vectors for foreign interference.