China’s State Security Minister Identifies US AI Models as Cyber Threats
China’s minister of state security, Chen Yixin, has named two American AI models—Claude Mythos and GPT-5.5-Cyber—as potential threats to China’s infrastructure in an official journal, China Cyberspace.
In an article published on Sunday, Chen outlines six risks associated with AI, emphasizing the models’ roles in raising the efficiency of vulnerability discovery and malware development. While he provides no specific examples of attacks, Chen warns that these AI advancements pose serious risks to China’s critical information infrastructure.
Key Points:
-
AI Risks: Chen identifies six risks, including political security threats, disrupted cyber offense and defense, amplified leak and espionage risks, AI monopoly-driven technological imbalances, structural shocks to social governance, and fundamental shifts in military conflict.
-
Political Rumors and Cognitive Warfare: He highlights the abuse of generative tools for creating political rumors and spreading harmful information, which can threaten political, institutional, and ideological security.
-
Data Leaks and Espionage: Chen expresses concerns about both Chinese users and foreign intelligence agencies accessing sensitive data, trade secrets, or citizens’ personal information through AI services.
-
Cyber Offense and Defense: The article emphasizes the transition to a new stage in cyber offense and defense characterized by industrialized vulnerabilities, fully automated attacks and defenses, and AI-against-AI battles, significantly lowering the barriers and costs of cyberattacks.
-
Model Names and Capabilities: While the article provides no examples of attacks, Claude Mythos and GPT-5.5-Cyber are cited as markers of AI’s growing capabilities in these areas.
-
Defensive Applications: Both Anthropic (developer of Claude Mythos) and OpenAI (developer of GPT-5.5-Cyber) market their models as security-focused, emphasizing their potential to aid in defense rather than conceal vulnerabilities.