When Anthropic's chief executive published a weekend essay advocating for measured progress in artificial intelligence development, China's government apparatus mobilized swiftly. The commerce ministry issued a statement characterizing the piece as evidence of American technological dominance ambitions, remarks that appeared in the Global Times on Monday. The state-aligned newspaper then commissioned a full investigation, bringing together three prominent researchers to dissect what they saw as contradictions in Amodei's position.
Amodei's Saturday post made two central arguments: that the AI sector should adopt a deliberate pace, and that Washington should "not sell powerful AI chips or semiconductor manufacturing equipment to China" while also preventing chip smuggling and restricting remote access to overseas data centres. The combination of these positions—calling for industry-wide restraint while simultaneously proposing restrictions on Chinese capabilities—became the focal point of Beijing's response.
Xiang Ligang, an observer of telecommunications and technology policy, characterized the proposal as "inappropriate, groundless and hostile." He argued that Amodei's framework contained an inherent tension: accepting that genuine AI pacing would eventually demand Chinese cooperation while simultaneously advocating export limitations and computing restrictions designed to slow Chinese advancement and preserve American technological superiority.
Xiao Qian, affiliated with Tsinghua's Institute for AI International Governance, highlighted a commercial dimension to the dispute, noting that Anthropic competes directly with Chinese open-source models. Liu Shaoshan, who leads embodied AI research at the Shenzhen Institute, offered a more expansive critique. Once artificial intelligence becomes framed as a technology determining national security and human civilization's trajectory, he contended, an AI company transcends its status as a conventional technology enterprise and functions instead as "strategic infrastructure."
A separate Global Times opinion piece, distributed through WeChat under the title "Targeting China's AI: America's 'tech right' rolls out a Cold War script," characterized Amodei's plan as a "quiet AI Cold War" and dismissed it as both hypocritical and strategically shortsighted.
What Amodei Actually Proposed
The Anthropic chief's essay outlined three distinct components: independent evaluators with employee-equivalent access to systems, which Anthropic had already committed to implementing; shared safety standards and capability thresholds across frontier laboratories operating in democratic nations; and a longer-term initiative to coordinate development pacing with non-democratic governments.
The first two proposals drew immediate support from industry peers, including OpenAI's Sam Altman, who endorsed them the same day. The export-control measures, however, occupied the same document and proved to be the element that captured Beijing's attention.
Escalating Tensions Over AI Exports
The two governments have been engaged in this dispute for weeks. On 9 September, the NSA, FBI and CISA publicly identified six Chinese firms engaged in industrial-scale model distillation, a technique for extracting capabilities from larger AI systems. Beijing rejected these allegations as baseless the following day.
In June, Chinese foreign minister Wang Yi proposed establishing a worldwide AI cooperation framework that would distribute affordable models to developing nations, a move that contrasted with Group of Seven discussions about restricting access to American AI systems. Days after that proposal, a Public First survey encompassing 18,000 respondents across 15 countries revealed that majorities in most nations already believed China had achieved technological superiority in AI capabilities.
What distinguishes this latest exchange is the target of Beijing's criticism. For months, the Chinese government has contested export restrictions as matters of state policy. This time, the commerce ministry directed its standard objections toward a private company's leader, based on a personal blog post he had authored just two days after calling on his own industry to embrace restraint. Meanwhile, former president Trump dismissed Amodei's slowdown argument, asserting that American victory in AI competition must take priority.
Source: The Next Web
