China’s Spy Chief Warns AI Could Fuel Political Rumors and Cyber Threats

China’s State Security Minister Chen Yixin has presented artificial intelligence as more than a technology-policy challenge. In his account, AI can be used to undermine political stability, target critical infrastructure and deepen China’s exposure to foreign technological leverage. 16

The core warning: AI as a political-security tool

Chen said “hostile forces” and people with ulterior motives could abuse generative AI to fabricate political rumors, spread harmful information and incite confrontational sentiment. He described these activities as public-opinion and cognitive warfare directed at China, with consequences for its political, institutional and ideological security. 4611

The concern is fundamentally about scale. Synthetic text, images, audio and video can make deceptive material cheaper to produce and easier to tailor for particular audiences. In the minister’s framing, that raises the risk that influence operations could be run at volume and used to disrupt social order or confidence in institutions. 24

This emphasis also aligns with China’s ongoing enforcement campaign against AI-generated misinformation, impersonation, violent or vulgar material, and other harmful content. The Cyberspace Administration of China has said its 2026 campaign also targets deficiencies in model registration and safety review, unsafe training data, data poisoning and failures to label synthetic content. 17

Why Chen singled out U.S. models

Chen’s article reportedly named Anthropic’s Claude Mythos and OpenAI’s GPT-5.5-Cyber in discussing dangers to China’s critical information infrastructure. Reporting on the Chinese-language text says the warning connected those models to increased efficiency in vulnerability discovery and malware development. 7

That is an assertion about potential capability and misuse risk—not evidence that the companies or models carried out an attack. One report specifically noted that Chen did not accuse either model of being used in attacks against China. 11

The broader security argument extends beyond disinformation. Chen warned of AI-enabled risks to critical infrastructure, data protection and warfare, while other reporting described concerns involving large-scale data breaches and technological monopolization. 912 In practical terms, Chinese officials are treating powerful foreign AI systems as possible tools that could aid cyber operations, automated collection, analysis or profiling if used by adversaries.

Technological self-reliance is part of the response

Chen’s proposed response was not limited to content moderation or cyber defense. He called for stronger investment in Chinese AI companies, computing resources, data and advanced chips so that China can independently control core technologies. 13

That framing turns AI supply chains into a national-security issue. Dependence on foreign models, cloud services or advanced semiconductors can be portrayed as a strategic vulnerability, especially amid U.S. restrictions on sales of advanced chips and chipmaking equipment to China. Chen also characterized AI as a new arena of major-power competition. 3

Beijing’s dispute with U.S. AI-slowdown proposals

The security warning emerged alongside a public clash over how AI safety should be governed. Anthropic CEO Dario Amodei called for a slower pace of frontier-AI development and argued for maintaining restrictions on China’s access to the most advanced chips and chipmaking equipment.

China’s state-backed Global Times cast that proposal as a “Cold War playbook” rather than a neutral safety agenda. China’s Foreign Ministry also criticized what it called fearmongering, confrontation and vicious competition.

This disagreement matters because both sides use the language of AI safety while emphasizing different dangers. Beijing’s public case combines misuse risks with political stability, data security and access to technology. U.S. debate, as reflected in the cited proposals and official reactions, is also bound up with maintaining an advantage over China and managing powerful-model risks.

What it means for U.S.-China AI talks

AI dialogue was reported as a desired Chinese deliverable ahead of the September 24 Trump-Xi summit, though a White House official said there was no planned AI-related meeting in mid-September. President Donald Trump has meanwhile played down demands for stronger AI safeguards, arguing that the United States must preserve its lead over China.

Those positions make an ambitious shared frontier-AI slowdown unlikely. The more plausible area for engagement is narrower risk management: discussion of cyber misuse, synthetic-media harms, incident communication or protections for critical infrastructure. Even that would face major obstacles, including export controls, access to advanced chips, data sovereignty and disagreement over what counts as genuine AI safety rather than strategic competition.

Chen’s intervention therefore signals a durable shift in tone: China is treating AI governance as inseparable from regime security and technological independence. That makes cooperation possible on specific harms, but much harder on rules that either side believes could constrain its strategic position.