▲ US and Chinese flags (file photo)
While the concept of human extinction and the need to regulate the pace of development have rapidly gained traction in the US artificial intelligence (AI) industry, government-level warnings regarding the risks of uncontrollable AI are also mounting in China.
These AI-related issues are drawing even more attention as they are expected to be a major agenda item during the upcoming summit later this month between US President Donald Trump and Chinese President Xi Jinping.
However, whereas discussions over AI potentially threatening human existence have begun in earnest in the US, the prevailing atmosphere in China is to acknowledge the risks of AI while keeping them at a manageable level, placing heavier weight on narrowing the technological gap with the US.
China has long issued state-level warnings about AI risks and has continuously updated and re-released guidelines based on AI development patterns every year.
The China National Cybersecurity Standardization Technical Committee raised the possibility that humans could lose control over AI when it released the "AI Safety Governance Framework 1.0" in September 2024.
The Chinese government stated through future scenarios that it cannot rule out the risk of AI autonomously securing external resources, self-replicating, pursuing external power, and potentially competing with humans.
In the version released last year (2.0), the risk scenarios became much more specific.
It assumed that before AI secures resources, self-replicates, or pursues power, its intelligence could leap forward unexpectedly and develop beyond anticipated levels.
This served as a more explicit warning about the possibility of AI slipping out of human control.
Consequently, Governance Framework 2.0 also added principles such as "using reliable AI" and "preventing the loss of control over AI."
Subsequently, in the 2026 version (3.0) released on the 14th, an analysis was raised that AI's capabilities to perform tasks on behalf of humans, interact with the physical world, and engage in deceptive behavior are continuously growing.
Concerns that AI might act contrary to human intentions were further highlighted.
These guidelines from the Chinese government go beyond simple warnings about AI risks, extending into a national security response stance.
Chen Yixin, Minister of China's Ministry of State Security—equivalent to South Korea's National Intelligence Service—stated in a recently published article that "AI has already become a major battlefield for global science and technology competition and a new race track for strategic games among major powers." He added, "We must firmly uphold a holistic approach to national security and accelerate the establishment of an AI security risk prevention system."
Minister Chen emphasized, "We must fully understand the risks and challenges brought by AI development," noting that "AI is penetrating every sector of human society at an unprecedented speed, scale, and depth."
Although no specific country was named, it is widely interpreted as largely targeting the US.
Minister Chen's remarks reveal a similar sense of urgency regarding the fear of losing control—which has recently spread across the US AI industry, spearheaded by Anthropic CEO Dario Amodei and others—while also showing that Chinese authorities are approaching AI risks as a national security issue.
President Xi also stated at the 2026 World Artificial Intelligence Conference (WAIC) held last July, "We must ensure that AI is always under human control." At the same time, he urged, "We must jointly oppose overstretching the concept of national security in the field of AI security and prioritizing one country's security over that of another."
Even though China has acknowledged these risks early on, the prevailing logic remains that it must first catch up with the US gap.
The Wall Street Journal (WSJ) summarized the Chinese industry's attitude toward the emergence of AI risks as "worries for developed nations."
Some engineers working at China's top AI research institutes reportedly feel that concerns over human doom caused by AI are problems belonging to "Western developed countries."
Although Chinese AI models often appear to surpass US models in global rankings, Chinese models are still considered to lag behind US models by several months or more.
In particular, because the technological advancement of AI is still in its early stages, risks threatening human existence feel like worries for the distant future.
Because of this, despite the awareness of AI risks, all government agencies and AI research institutes in China prioritize technological advancement above all else.
Major Chinese AI companies, including DeepSeek and Zhipu (Z.ai), are pushing forward with the development of Artificial General Intelligence (AGI), where AI thinks like humans.
Some analyses also suggest that a culture of evaluating and reflecting on the social impact of technological advancement is less established in China.
Zilan Chen, a researcher at the Oxford China Centre, pointed out to the WSJ, "Because AI development is happening faster in the US and American technology is more cutting-edge, people in US labs may be more attuned to the risks than those in Chinese labs."
(Photo: AP, Yonhap News)
※
Copying, redistribution, and unauthorized use in AI training are strictly prohibited.