China has strongly pushed back against Anthropic CEO Dario Amodei’s controversial proposal to reshape the global artificial intelligence landscape, dismissing the call to slow down frontier AI development as another veiled attempt to preserve American technological dominance.
Earlier this month, Amodei published a widely discussed blog post arguing that the international AI industry must intentionally slow the training and deployment of its most powerful models. According to the chief executive of the prominent U.S. artificial intelligence lab, implementing a deliberate pause would provide governments and researchers vital breathing room to establish safety protocols, mitigate catastrophic risks, and manage the rapid ascent of autonomous systems. Crucially, Amodei acknowledged that any effective pacing strategy would "require cooperation with China, the autocratic country with by far the most advanced AI capabilities."
To achieve this, Amodei outlined a series of strict enforcement measures designed to deliberately stall China’s artificial intelligence trajectory. His blueprint called for a complete cessation of sales regarding advanced AI accelerators and semiconductor manufacturing equipment to Chinese entities, a comprehensive international crackdown on high-end chip smuggling networks, and strict legal and technical restrictions preventing remote access to foreign data centers located outside of mainland China.
The proposal exposed deep ideological and strategic fractures within the Western technology sector. Amodei secured unexpected backing from several key U.S. competitors, including OpenAI CEO Sam Altman and xAI founder Elon Musk, both of whom have frequently warned about the existential hazards posed by uncontrolled artificial intelligence systems. Conversely, Meta CEO Mark Zuckerberg and President Donald Trump outright rejected the concept, instead advocating for rapid, unfettered technological growth to maintain national competitiveness.
Beijing’s response, however, has been swift and uncompromising. Chinese government officials, state-run media outlets, and leading industry engineers have mobilized to reject a framework they view as an exclusionary geopolitical maneuver designed to lock China out of the highest echelons of technological progress.
"I don’t see any reason why China would cooperate with Amodei’s proposal and be held down more than it already is," said Alvin Wang Graylin, a senior fellow for technology at the Asia Society Policy Institute’s Center for China Analysis. Graylin noted that while Beijing might eventually be open to international dialogue regarding the regulation and governance of artificial intelligence, it would never agree to abide by conditions dictated entirely by American industry leaders under Washington’s terms.
The friction over artificial intelligence policy comes at a delicate time for bilateral diplomacy. U.S. President Donald Trump is scheduled to meet with Chinese President Xi Jinping in Washington later this month, with American officials widely expected to raise the issue of "shared risks" stemming from advanced foundational models during the high-level discussions.
Cold War Playbook?
Beijing wasted little time framing Amodei’s proposal within the context of intensifying geopolitical rivalry. Just one day after the publication of the Anthropic executive’s blog post, the state-backed Global Times published a sharp editorial titled "Targeting China’s AI: U.S. ‘tech right’ unfolds Cold War playbook." The publication argued that Amodei’s suggested restrictions on hardware and Chinese AI enterprises were transparently protectionist measures engineered to safeguard American technological hegemony under the guise of existential safety.
On September 11, Chinese Foreign Ministry spokesperson Guo Jiakun formally condemned the initiative, describing Amodei’s warnings as little more than "fear-mongering, confrontation, and vicious competition" that would ultimately fracture rather than improve global AI governance.
Despite rejecting the U.S. CEO’s specific stipulations, analysts point out that China possesses its own increasingly sophisticated and expansive framework for evaluating artificial intelligence safety. On September 14, China’s national cybersecurity standards body released the third edition of its Safety Governance Framework. The regulatory document explicitly warns that advanced artificial intelligence systems could autonomously learn, optimize themselves, and undergo "recursive self-improvement" at a velocity or along a trajectory that rapidly exceeds human expectations and control.
This domestic policy concern closely mirrors dystopian scenarios voiced by Western researchers, including former Anthropic safety scientist Jacob Coxon, who warned earlier this month about the severe vulnerabilities associated with advanced autonomous reasoning.
"I do not think China and the U.S. see AI risks in fundamentally different ways," explained Zilan Qian, a London-based research scholar at the artificial intelligence governance think tank GovAI. "But their socioeconomic conditions and political institutions make some risks more visible and some voices more prominent in each country."
Is China Worried About AI Dystopia?
According to Qian, Chinese policymakers are deeply concerned about a wide array of artificial intelligence risks, including the destabilizing influence of synthetic media on the national information ecosystem, the psychological and societal impacts of citizens forming romantic relationships with conversational chatbots, potential systemic disruptions to the domestic labor market, and the exploitation of automated tools by malicious actors to execute sophisticated cyberattacks.
"Compared with U.S. frontier labs, there is less discussion in China of frontier AI risks, such as humans losing control of AI or AI automating AI research," Qian observed. "However, these risks are increasingly being acknowledged by Chinese labs and in policy documents."
The security implications of foundational models have also drawn direct attention from China’s national security apparatus. Days before Amodei published his essay, Chinese State Security Minister Chen Yixin warned that advanced American models, such as Anthropic’s proprietary systems and rival architectures, posed severe threats to China’s critical information infrastructure and national security apparatus.
Chen urged domestic authorities to implement much tighter government oversight, accelerate efforts toward absolute technological self-reliance, and build robust defenses against AI-enabled offensive cyber operations and foreign influence campaigns.
"Modern warfare has entered an era increasingly shaped by artificial intelligence," Chen wrote in a contribution to the state-run magazine China Cyberspace. "Whoever can use algorithms to achieve more refined sensing, more accurate decision-making, and more precise strikes will gain the initiative on the battlefield."
Who Defines the Game?
While China’s corporate leaders and major artificial intelligence conglomerates have largely maintained an official silence regarding Amodei’s proposal, engineers and researchers actively building these complex systems have been far more vocal in their criticism.
Tao He, head of research and development at MiniMax Code, a prominent Chinese frontier AI lab, took to social media to voice his frustration with the shifting narrative coming out of Silicon Valley.
"Oh, come on. Yesterday, the mission was AGI. Today, we’re told to slow down," He wrote on the platform X. "Why do you always get to define the game—and change the rules?"
Similarly, DeepSeek engineer Liu Shengyu, who contributed to the development of the firm’s advanced models, articulated his own perspective on the risks of technological centralization in a widely shared personal essay published on WeChat. Liu warned of a future where a small handful of American technology conglomerates maintain a monopoly over society’s most powerful artificial intelligence models, thereby consolidating an overwhelming share of global economic and technological power.
Liu argued that entities holding a monopoly on frontier intelligence would become virtually impossible for international competitors to challenge, locking the rest of the world into a permanent cycle of social and economic disadvantage. To counteract this outcome, he advocated for making powerful artificial intelligence models inexpensive, open, and widely accessible.
"I don’t trust Anthropic or OpenAI to do that," Liu wrote. "In particular, I do not want Anthropic to control the world’s most advanced AI or AGI."
For Chinese developers who are already operating under stringent U.S. export controls and hardware limitations, calls from Western executives to decelerate research carry an entirely different weight, particularly while their domestic models are still striving to close the performance gap with American industry leaders.
Addressing these disparities last week, Huawei rotating chairman Eric Xu suggested that leading U.S. laboratories may be encountering systemic risks that Chinese developers are not yet powerful enough to perceive. Rather than putting a brake on innovation, Xu argued that Chinese researchers must accelerate their development cycles to reach a threshold where they can independently evaluate and experience these advanced frontier risks firsthand.
"I think maybe AI model providers in China may need to speed up their pace to the level that they could also feel the risks from AI development," Xu said.
