China is reacting strongly against Dario Amodei, CEO of Anthropic, following his comments advocating for a slowdown in AI development. Amodei identified China as a major risk factor in AI safety and called for increased regulations targeting the country. In response, Chinese experts criticized his remarks, stating that they frame AI safety issues as a geopolitical zero-sum game. As the US-China competition in AI intensifies, there are concerns that the upcoming summit on September 24 may yield limited results due to the differing positions of the two nations.
CEO Amodei emphasizes need for regulations on China in AI development
In a blog post on September 12, Amodei mentioned China 14 times, arguing that the US should slow its AI development only if it remains ahead of China. He warned that China's advancements in AI could pose significant security threats to the US and the world. He advocated for stricter export controls on high-performance AI chips and semiconductor equipment to China, as well as measures to prevent chip smuggling and unauthorized access to remote data centers. He also stressed the need to block unauthorized distillation of US AI models by Chinese companies.
Amodei argued that these actions could significantly enhance the US's AI advantage over the next three to five years, thereby increasing its negotiating power with China. He contended that if the US were to slow its AI development first, it could risk ceding technological superiority to China, making it imperative to curb China's advancements for the sake of international AI safety cooperation.
China criticizes AI safety claims as zero-sum competition
Chinese officials quickly condemned Amodei's statements. Xiang Ligang, a telecommunications and technology policy expert, told the state-run Global Times that Amodei's claims depict China's legitimate AI development as a threat and further fuel the US-China AI rivalry. He described the assertions as inappropriate, unfounded, and hostile. Xiao Qian, deputy director of Tsinghua University's AI International Governance Research Institute, noted that Anthropic's push for regulations stems from its competition with China's open-source AI models.
On September 13, Chen Yixin, head of China's Ministry of State Security, issued a warning that some countries are using 'national security' as a pretext to impose technology controls and monopolize industrial standards, thereby fragmenting the global AI industry and supply chain. While he did not name specific countries, his remarks were interpreted as a direct reference to US regulations targeting China.
China is also seeking to expand its global influence by promoting open-source AI models in contrast to the closed AI models led by US tech giants. During the BRICS summit on September 12, President Xi Jinping proposed creating a dedicated zone for open-source AI for BRICS nations (Brazil, Russia, India, China, and South Africa) and supporting collaboration in the development and application of large language models (LLMs) to establish an open AI ecosystem.
Institutional limitations complicate cooperation amid 'prisoner's dilemma'
While both the US and China agree on the necessity of AI safety, they hold opposing views on how to achieve it. The US perceives China's AI advancements as a security risk and seeks to limit technology access, whereas China criticizes US technology controls as actions that lead to monopolization and supply chain disruptions.
The competitive nature surrounding AI development further complicates cooperation between the two nations. Although collaboration could mitigate AI risks, mutual distrust prevents either side from halting their AI development efforts, resulting in a 'prisoner's dilemma.'
Former President Donald Trump acknowledged the need for AI safety measures while emphasizing the importance of maintaining a competitive edge over China, stating, 'The one who wins in AI will win.' Although the US and China are set to hold their first official AI safety dialogue this month, analysts predict that reaching a substantive agreement will be challenging.
In addition to the competition for technological supremacy and mutual distrust, institutional limitations pose further obstacles. A recent report from the Asia Society, a US think tank, highlighted that both the US and China have dispersed technical expertise and policy authority across multiple agencies in the field of AI safety, making it unclear who would be responsible for implementing any substantial agreements.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.