The head of China's top intelligence agency has publicly warned that artificial intelligence could directly threaten the Chinese Communist Party's grip on power and the country's political stability. The stance contrasts with growing concern in the United States that AI could escape human control and endanger humanity as a whole, as Beijing frames the technology as a matter of regime security and national security.
Chen Yixin, minister of China's Ministry of State Security, argued in a recent article for the state-run magazine China Cyberspace that Communist Party control and government oversight of AI must be tightened further, the New York Times reported on the 14th.
He identified maintaining political stability, defending against cyberattacks and disinformation campaigns by foreign forces, and military competition with the United States and others as the central tasks of AI management.
The first risk Chen raised was the stability of Communist Party rule. He warned that forces with malicious intent could use AI to produce fake video and audio and wage propaganda campaigns targeting China. A deepfake showing Chinese President Xi Jinping or a senior party official saying something they never said could spread instantly and lead to social turmoil, he said. Given that the Chinese government has long treated regime stability and the preservation of Communist Party rule as its top priorities, Beijing has effectively begun to define AI not as a purely technological risk but as a threat to regime security.
Chen also emphasized cybersecurity risks. Citing specific systems developed by U.S. technology companies, including Claude Mythos and GPT-5.5-Cyber, he said AI is sharply lowering the technical and economic barriers to mounting cyberattacks. He also raised the possibility that foreign intelligence agencies could use AI to conduct large-scale espionage against China or attack key information and telecommunications infrastructure.
The leakage of sensitive information through the use of foreign AI models is another Chinese concern. Anthropic of the United States recently claimed that China's Moonshot AI had rerouted some user queries to Anthropic's own AI, Claude. Anthropic said surveillance footage linked to the Chinese military and sensitive information from Chinese technology companies were transferred to its systems in the process.
Chen said AI is also fundamentally changing the nature of warfare. Citing the use of AI by the United States and Israel in their conflict with Iran, he predicted that countries able to use algorithms to detect, identify and strike enemies faster and more accurately would seize the initiative on the battlefield.
Chen's argument is seen as the most comprehensive account of AI risk the Ministry of State Security has put forward to date. Xi's leadership is pursuing a dual strategy of cultivation and control, nurturing AI as a strategic industry that China must secure in its technological rivalry with the United States while tightening political and social controls.
That approach differs sharply from the AI safety debate now unfolding in the United States. Anthropic Chief Executive Dario Amodei, OpenAI CEO Sam Altman and Google DeepMind Chief Executive Demis Hassabis have raised the possibility that AI could surpass humans' ability to control it and called for slowing the pace of development. China's concerns, by contrast, center on the possibility that AI could shake Communist Party rule or be used as a foreign tool for information warfare and military operations.
The gap in perceptions between Washington and Beijing is expected to shape future AI safety talks between the two countries. With AI-related discussions anticipated ahead of a summit between President Donald Trump and Xi in Washington later this month, analysts say China is likely to present data sovereignty and political security as its core red lines in the negotiations.
In the end, facing the same set of AI risks, the United States worries about a loss of control — that AI could surpass humans — while China sees a more direct threat to regime security, that AI could loosen the party's and the state's grip.
