President Donald Trump has stated that the United States will not cooperate with China on artificial intelligence (AI) management. In contrast, China argues that a common understanding is necessary for managing AI risks.
Last week, the U.S. and China established a dialogue on AI risk management and agreed to hold follow-up discussions in November. However, differing approaches to AI safety suggest that concrete cooperation may be challenging.
According to reports, Trump drew a line on cooperation with China regarding AI risk management during a government website launch event in Washington on September 29.
He claimed that pursuing joint projects in AI with China would be akin to handing over U.S. secrets, making it difficult for American companies to maintain a competitive edge over their Chinese counterparts.
Trump has consistently argued for avoiding direct government regulation of AI to maintain an advantage in the competition with China, emphasizing the importance of self-regulation by companies.
During a lunch meeting at the White House with leaders from major U.S. tech companies, he reiterated the need for companies to autonomously address AI safety.
At this meeting, six AI companies (Google, Anthropic, Meta, OpenAI, xAI, and NVIDIA) signed a self-regulation agreement focused on AI safety. The agreement includes provisions for internal controls to prevent AI systems from unintentionally infiltrating or hacking other systems, as well as evaluations by independent external auditors to ensure these controls function properly.
Conversely, China maintains that cooperation between the U.S. and China is essential for addressing AI safety issues. Chinese President Xi Jinping emphasized the need for both countries to share responsibility for AI development and management during a recent meeting with Trump.
During the U.S.-China summit, both nations exchanged views on the risks and benefits of AI and agreed to establish communication channels for addressing AI-related incidents. They also committed to continuing the AI dialogue in November.
Following the summit, a Chinese think tank argued that the U.S. and China should develop common standards and terminology for managing AI risks.
According to the South China Morning Post, Liu Chong, director of the Science and Technology Strategy Research Center at the China Institute of Contemporary International Relations (CICIR), stated that the focus of U.S.-China AI dialogue should be on forming a common understanding of AI risks and building mutual trust, rather than solely addressing technological competition.
Liu noted that the two countries do not need to adopt the same AI industrial policies or share core information and technologies. Instead, he suggested that governments, companies, and researchers should first establish common terminology and understanding of AI risks for discussion. He also proposed creating a system to adjust responses based on changes in risk levels.
He argued for increased oversight of companies developing advanced AI, stating that a few leading firms should not dictate the direction of AI management. He highlighted the difficulty of assessing risks associated with closed AI models and called for U.S. companies to be more proactive in disclosing information related to AI risks.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.