China spy chief warns AI is a ‘new arena for strategic rivalry,’ SCMP and Bloomberg report

China’s state security minister Chen Yixin wrote in the Cyberspace Administration journal China Cyberspace that artificial intelligence has become “a new arena for strategic rivalry among major powers,” warning of political and cyber risks and citing next-generation U.S. models such as Anthropic’s Claude Mythos and OpenAI’s GPT-5.5-Cyber. South China Morning Post and Bloomberg carried English wraps Monday; Caixin printed the Chinese signed essay.

While U.S. lab CEOs argue about pacing the frontier, Beijing’s top intelligence official frames AI as great-power rivalry and names specific Western cyber-capable models as threats — a different China signal from Global Times slamming Amodei’s essay, and relevant to any U.S.–China AI safety talks.

Chen Yixin, China’s state security minister — the head of the Ministry of State Security (MSS), the country’s main civilian intelligence service — published a signed essay in China Cyberspace, a journal of the Cyberspace Administration of China (CAC), the internet regulator. The journal posted it on its social account Sunday. South China Morning Post’s Carol Yang, writing from Beijing, and Bloomberg carried English wraps Monday. Caixin printed the Chinese text in full on 13 Sep at 22:59. That signed essay is the filing event. This desk does not treat it as a Ministry of Foreign Affairs press conference.

The quoted line, as SCMP reports Chen: artificial intelligence has become “a new arena for strategic rivalry among major powers.” File those as Chen’s words via the essay and the English wraps. This desk is not endorsing the frame.

Chen warned of a systemic impact on ideological security — the official term for protecting the Party’s political narrative. He said hostile forces could use deepfakes and AI spam for cognitive warfare: fake audio and video, plus mass-produced junk posts meant to shape what people believe. Bloomberg quotes him saying abusive use of AI by adversaries could “directly threaten China’s political security, institutional security and ideological security.”

He named two next-generation U.S. models — Anthropic’s Claude Mythos and OpenAI’s GPT-5.5-Cyber — as lowering the technical threshold and costs of cyberattacks against critical infrastructure. File the names as Chen’s examples. This desk is not confirming a new attack, and it is not saying those models were banned.

The essay lists six risk areas, in plain English: political and ideological security; cyber offense and defense; mass data leaks when people put sensitive files into foreign AI products; tech monopoly and export controls that split supply chains; a shock to social governance — how the state keeps everyday order; and military AI becoming a key battlefield role, not just a helper.

The essay’s title, as Caixin reprints it, is “Fully build an AI safety barrier and promote healthy, orderly AI development.” File the title. Do not read it as a ban on named models or a new export rule.

CONFIRMED here: Chen Yixin’s signed essay as reprinted in full by Caixin from China Cyberspace. REPORTED: SCMP and Bloomberg English wraps. NOT claimed: that Chen banned models, announced new export rules, or held a press conference. Distinct from the already-filed china-gt-amodei-cold-war, xi-brics-ai-open-source-community, amodei-cbs-china-toughest-dilemma, amodei-pace-the-frontier, and nsa-ai-mission-center.