AI Photo: VCG
Speaking at a hearing of the US Senate subcommittee, White House officials proposed blocking the development of artificial intelligence (AI) in China by exporting “AI technology stacks” that integrate not only chips but also algorithms and applications. Meanwhile, the US Senator defended the US global AI domination and claimed that he would win the AI competition at the same hearing, according to the US Senate Committee, Science and Transportation Agency website.
Chinese experts dismissed the idea as “hopeful thinking rooted in static logic.” This approach is myopic, counterproductive to innovation and counter-spirits of the collaborative spirit necessary for global advancement in AI.
At a local Senate Subcommittee hearing on Wednesday, White House Science and Technology Policy Director Michael Krazios said in 2020 that American Innovation Enterprises gained a more comfortable lead than their closest competitors in AI, but by 2024 the gap was beginning to get significantly closer. “We were at risk of losing excellence in this critical technology,” Kratzios said.
“It's a moment when we actually have competitive technologies. We have the best chips, models and applications of the US government to help promote these technologies widely.
“Simply put, Kratsoos, who represents several US voices, wants to strengthen their AI rivalry. They sell premium AI technology stacks to allies, but not China, but the US dominates the market, preventing China from moving forward, and ensuring US leadership forever. This mindset ignores long-term outcomes.
Meanwhile, in an age where AI promises are collectively progressing, Zhang said Kratsios' zero-sum games risk destroying the open ecosystem that defines field development.
According to media reports, Tech Stack is a collection of technologies that include advanced semiconductor chips and algorithms used to develop applications. This was cited in the US AI Action Plan announced in July. This states that the US will “provide secure, full-stack AI export packages to American friends and allies, including hardware, models, software, applications, and standards.” The Action Plan also argued that it would counter China's influence on international governance agencies.
Apart from targeting China, experts are using US AI technology and chips to doubt the security of American allies. This is as Kratsios argued at the hearing that “if not the norms around the world, those models and their applications will be fine-tuned with enemy models running on hostile chips.”
Analysts pointed to potential security loopholes for US chips. Xinhua's news agency reported in August that for months, some U.S. lawmakers have been pushing for months to install “hardware kill switches” and essential tracking devices on AI chips through laws such as the so-called Chip security law.
The bill could embed location trackers in manufacturers into chips, monitor international movements, and allow remote disabling without user consent, Xinhua said.
Zhang said that in terms of the “effectiveness” of this only existing approach, Kratzios' claim that “exports the AI stack to allies while maintaining American leadership while limiting China and maintaining American leadership” would simplify the global AI ecosystem. AI thrives with open collaboration, data sharing and diverse contributions.
“The advancement in AI in China, driven by autonomous innovation and open collaboration, shows resilience to containment,” Zhang said.
Official data shows that China has more than 4,500 AI-related companies live in China by May, with the AI industry worth nearly 600 billion yuan. In 2024, Chinese companies accounted for more than 61% of the world's AI patent applications, while new industries, business formats and digital models contributed to China's GDP of over 18%.
“Kratzios' strategy ignores this momentum, assuming that restrictions will curb China's growth. Instead, such measures will only accelerate China's independence,” Zhang said.
In some areas such as chip architecture and algorithms, China has achieved breakthroughs and is building an independent ecosystem spanning AI chip R&D, algorithm development, data annotation, model training and applications. This integrated chain promotes synergies and reduces costs, Zhang pointed out.
In particular, in response to the rapid rise of China's open source AI models, the Trump administration has shifted from the Biden administration's concern that the open source model could allow its competitors to gain US technical advantages. Instead, they have actively supported open source development, particularly to counter China's growing open source model and the possibility of setting global standards. Xiao Qian, dean of the Department of International Affairs, Tsinghua University, said in an article published Tuesday on the China-US Focus Platform.
On August 5th, Openai released its first open source model in five years, released the GPT-Oss, and entered the competitive field of the open force ecosystem. Commentators suggest that the move is aimed at reasserting open leadership in the open source domain, and is directly challenging China's advances in open source AI, Xiao said.
Chinese experts also highlighted the fear-driven strategy of Kratsooth is at risk of fragmented AI landscapes, stifling innovation and economic growth. Instead of containment, the US should embrace competition as a catalyst for progress and work with China and others to set global AI standards. This ensures technological advancements, economic stability and shared prosperity, consistent with the open mind spirit necessary for AI's potential.
When asked to comment on the US AI Action Plan, which argued that “strengthen US control before China,” China said it “believes that all parties should jointly promote the openness and comprehensive development of AI. Achieve a common development.
