Adversary Intelligence Services — 2026-09-15

China Spy Chief Chen Yixin Warns AI Poses Direct Threat to Political Security and Names Western AI Models as Hacking Risks

BLUFChen's public naming of specific Western AI models as security threats is unlikely to produce formal restrictions within 30 days, serving instead as ideological positioning ahead of the Xi‑Trump summit.

China's Minister of State Security Chen Yixin wrote in an article published online on September 13 that "hostile forces" are abusing generative AI to "fabricate political rumours and spread harmful information," warning this constitutes "public opinion wars and cognitive warfare" that directly threatens China's political, institutional, and ideological security 123. Chen named Anthropic's Claude Mythos and OpenAI's ChatGPT-5.5-Cyber specifically, citing their hacking capabilities as posing "serious risks to China's critical information infrastructure" 12. He also said foreign intelligence agencies use AI-driven web crawling, data mining, and profiling to collect Chinese state data, trade secrets, and citizens' personal information, and called for China to secure "independent control" of core technologies including advanced chips 123. The article's publication precedes a planned meeting this month between Xi Jinping and US President Donald Trump in Washington, at which AI governance is expected to be on the agenda 123.

Analysis
Naming Claude Mythos and ChatGPT-5.5-Cyber by name elevates specific Western models from generic "foreign technology" rhetoric to named national-security threats, positioning Beijing to justify future restrictions on their use in government and infrastructure systems. This marks an escalation from MSS's June warning on generic unauthorized AI relay services, now tied explicitly to chip sovereignty. A formal CAC or MSS notice restricting these models by mid-October is unlikely: the language reads as ideological positioning ahead of the Xi-Trump summit, with no CAC or MIIT drafting activity yet surfaced, and the rhetoric may instead function primarily as domestic messaging to harden China's negotiating posture. Confidence is low, resting on a single primary Chinese-language source (China Cyberspace) reproduced verbatim by multiple outlets via one AFP wire feed rather than independent reporting. Absent a restrictive notice, the article stands as negotiating leverage rather than an imminent compliance burden for Anthropic and OpenAI in the Chinese market.
5 sources
  1. China spy chief warns foreign hostile forces may use AI to fabricate political rumours - Hong Kong Free Press
  2. China's Spy Chief Warns of National Security Risks from AI - Asharq Al-Awsat English
  3. China's spy chief warns of national security risks from AI - The Standard (Hong Kong)
  4. China's Top Spy Chief Warns AI Is a Threat to Party Rule - GV Wire
  5. 国家安全部部长陈一新：全面筑牢人工智能安全屏障 推动人工智能健康有序发展 ("Comprehensively Building AI Security Barriers, Promoting Healthy and Orderly Development of AI") - China Cyberspace (中国网信), official WeChat account/journal of the Cyberspace Administration of China

View in full brief →

UNCLASSIFIED // OPEN SOURCE