AI-driven doomsday scenarios from US tech companies are seen in China as a largely Western concern. According to some experts, these warnings are perceived as a tactic to prevent Chinese artificial intelligence companies from surpassing their US competitors.
In Silicon Valley, software developers fear that an artificial intelligence too advanced for humans to control could start a global drone war or set off a nuclear bomb. In Beijing, pulling the plug, imposing an internet blackout and detaining those responsible in case artificial intelligence gets out of control is considered a definitive solution to the problems that artificial intelligence may cause.
US tech companies and officials have warned that artificial intelligence poses a catastrophic threat to humanity after major US laboratories revealed that advanced models had evaded testing protection barriers and infiltrated real-world systems. Leading US experts in the artificial intelligence sector called for Beijing and Washington to work together.
But for many in China, these apocalyptic warnings seem distant and distinctly Western. The idea that the ruling Communist Party could lose control over a technology is almost inconceivable. Authorities have previously successfully policed other potentially destabilizing technologies, particularly the internet. During the pandemic, the government had demonstrated its ability to contain major crises by locking down entire cities and tracking down hundreds of millions of residents.
'IT IS IMPOSSIBLE TO CONVINCE MANAGERS'
According to the New York Times (NYT), Afra Wang, who writes on artificial intelligence in Silicon Valley and technology in China, said, "Managers can instantly cut off the electricity grid of an entire city. They have tremendous power in the physical world."
Wang stated that it does not seem possible to convince the Chinese administration "why something like artificial intelligence living on a computer is scary." According to the expert, China focused on the dangers of artificial intelligence not on the risk of advanced models waging war on humans, but on the fact that traditional cyber security systems can be more attacked and the use of new technologies in malicious areas such as biological weapons studies.
"FEAR MARKETING OF THE USA"
In more skeptical interpretations, doomsday warnings are seen as a tactic to prevent Chinese artificial intelligence companies from surpassing their US rivals. According to reports in Chinese media, US artificial intelligence companies are engaging in "fear marketing" to maintain their place in the industry.
“In local industry circles or among the people around me, most people ignore the possibility of AI getting out of control,” said Jie Cai, who until recently researched human-computer interaction technology and AI safety at Tsinghua University.
Stating that the view that human beings are at existential risk seems a distant possibility for many people in the industry in China, Cai said, "We think that artificial intelligence technology is still at a very early stage. "The focus in the sector is still clearly on opportunity and innovation," he said.
'IT IS LIKE A FRANKENSTEIN NOVEL'
For some, apocalyptic narratives about artificial intelligence seem particularly foreign. A Chinese blogger compared Western fears of artificial intelligence to Mary Shelley's novel "Frankenstein" and noted that "this narrative tradition does not exist in Chinese culture."
Liang Fang, Head of AI Security and Governance at Beijing-based AI security consultancy Concordia, noted that Chinese culture emphasizes the unity of humanity and the idea that humans are an integral part of nature.
“Everything, including artificial intelligence, comes from nature,” Liang said. Therefore, artificial intelligence technology is seen as a companion that exists with us rather than an existential threat,” he said.
CHINA'S ARTIFICIAL INTELLIGENCE CONCERNS
The release of Mythos, a powerful Anthropic frontier AI model, and others like it have been met with concern in China. Authorities have issued new guidelines and warnings about artificial intelligence agents getting out of control, being exploited by hackers, or being used to make biological weapons.
Authorities have drafted a cybercrime law that would require AI companies to monitor and stop their models from being used to create malicious code that could be used in cyberattacks. Beijing administration also shared the updated version of China's "Artificial Intelligence Security Governance Framework" this month. The document reflected some Silicon Valley fears about “recursive self-improvement,” which is when AI systems become so powerful that they improve themselves without the help of human developers and spiral out of control.
Earlier this month, Wang Lihong, a senior official at the Internet regulator, the Cyberspace Administration of China, said that AI-supported cyber attacks are rendering traditional defense systems useless, adding, "A high level of vigilance against the risks of loss of control in AI is urgently needed."
RESEARCHERS HAVE WARNED FOR YEARS
Some Chinese researchers have been warning for years about the risks of advancing artificial intelligence too quickly. In 2024, Pan Xudong of Fudan University in Shanghai and other researchers discovered that two popular large language models from Meta and Alibaba showed the ability to replicate themselves. This was interpreted in the West as a step towards artificial intelligence being able to defeat humans.
In a speech at an AI safety conference last year, Pan said he and his co-authors "wanted to convey that such a risk is no longer limited to the realm of science fiction, but is a real problem that requires discussion and investigation."
In Beijing's eyes, artificial intelligence represents a power that can be used for the good of the country with careful planning. China has already enacted some of the world's strictest industry regulations: AI-generated content in China must be labeled; AI models must be registered and approved by authorities before they are released into the market.
While China has extensive regulatory and testing requirements, there is no legal requirement to test for catastrophic risks, such as using an AI system to make chemical or biological weapons.
AI COMPANIES ARE BEHIND IN TRANSPARENCY
Sarah Sun, founder of the Shanghai-based Open Community for AI Safety China, said the debate about AI threats in the US inspired the debate in China. However, Sun noted that Anthropic CEO Dario Amodei's criticism of China sparked a backlash. In his open letter about the dangers of artificial intelligence, Amodei argued that democracies should maintain their leadership in technology against authoritarian countries such as China. Amodei had wanted China's access to advanced chips and other equipment needed to develop artificial intelligence to be further restricted.
Such messages make recent AI warnings seem like justification for holding China back rather than genuine concern about the risk of disaster, Sun said. The expert added that this situation makes it difficult for people like him to raise awareness about artificial intelligence security in China.
REACTION TO ANTHROPIC IN CHINA
Online commenters in China called Anthropic "A-chu". It's a derogatory term that likens the AI giant to a monster. Yuyuan Tantian, a blog affiliated with China's state broadcaster, said the only thing US tech leaders like Amodei want to slow down is the speed at which others are catching up with them in artificial intelligence technology.
Zhou Hongyi, an influential Chinese cyber security manager, argued that the solution is not to slow down, but to accelerate defense capacity with artificial intelligence.
"Use AI to defeat AI. If you use AI to attack, I will use AI to defend," Zhou explained in a video posted online last week.
"CHINA DOES NOT TRUST THE USA, AND THE USA DOES NOT TRUST CHINA"
In an online post in early September, an engineer from Chinese AI startup DeepSeek noted that allowing Anthropic to monopolize the world's most advanced AI tools "would be no less serious than Hitler's first acquisition of atomic bomb technology in World War II."
Even though Chinese leader Xi Jinping told US President Donald Trump during his visit to New York for the UN General Assembly that they should "manage artificial intelligence for good", an agreement between the US and China on artificial intelligence security seems unlikely.
