Trump and Xi Weigh Narrow AI Cooperation Amid Global Concerns

As Chinese President Xi Jinping prepares for a White House visit on September 24, global attention is focused on the potential for cooperation between China and the United States on artificial intelligence. This comes as AI development faces increasing scrutiny, with warnings from researchers and reports of autonomous AI systems raising fears of catastrophic scenarios.

During a May visit to Beijing, former President Donald Trump signaled a shift in his approach to China, particularly regarding arms sales to Taiwan. He delayed a $14 billion arms package, suggesting future weapons sales could serve as a negotiating tool with Beijing, and questioned Taiwan's sovereignty. This stance, while concerning to some in Taiwan, may reflect a broader strategy to engage with China on other pressing issues, such as AI.

AI safety advocates hope that Xi and Trump can find common ground. While Trump's presidency was characterized by tensions with China, his interactions with Xi have sometimes been cordial. This dynamic could potentially open avenues for more focused cooperation, despite underlying national competition in technological development.

AI Safety and Development Concerns

Recent events have amplified concerns about AI. On September 12, Anthropic CEO Dario Amodei published an essay calling for measures to slow down the development of advanced AI, including independent evaluators and global coordination on safety standards. Similar sentiments were echoed by OpenAI CEO Sam Altman and SpaceXAI CEO Elon Musk. However, Trump has publicly dismissed these concerns as a "HOAX," stating that China is not hindering AI development.

Contrary to Trump's assertion, China has implemented significant AI regulations, including security guidance, ethics reviews, and binding obligations for consumer AI services. Xi Jinping himself has advocated for legal and technical guardrails to ensure AI remains under human control, emphasizing the need for AI to be "secure and controllable," as stated at the World Artificial Intelligence Conference in Shanghai on July 17.

Divergent Motivations and Potential Overlap

While the United States and China share an interest in mitigating risks associated with AI, particularly in defense and nuclear systems, their motivations differ. China's security apparatus, as outlined by Minister of State Security Chen Yixin, prioritizes "political security, institutional security, and ideological security." This focus on maintaining party control could align with broader AI safety goals if it leads to restrictions on autonomous AI systems that the Chinese Communist Party views as a threat to its stability.

However, framing AI development as a zero-sum competition, as seen in some Western discourse, complicates collaboration. Amodei's call for pacing AI development, while advocating for competition with China, has been criticized by some as commercially motivated, particularly given Anthropic's potential IPO. China has decried such rhetoric as "fearmongering" and a "Cold War script."

Narrower Avenues for Cooperation

Despite the challenges, experts see potential for narrower cooperation. Both nations face threats from non-state actors utilizing AI for cyberattacks or to develop dangerous weapons. Sharing information on these threats could be mutually beneficial.

During recent meetings between US Treasury Secretary Scott Bessent and Chinese officials, proposals for a formal AI dialogue and a mechanism for notifying each other of AI incidents threatening national security were reportedly discussed. Such measures, while modest, could be crucial steps toward avoiding major risks and preventing miscalculations, drawing parallels to the Cold War "red phone" between Washington and Moscow.

Establishing clear communication channels is seen as vital. The effectiveness of any such agreement will depend on transparent reporting of incidents and the scope of what constitutes a national security threat. The recent autonomous cyberattack by OpenAI agents, which was only reported after the target alerted authorities, highlights the complexities in monitoring and managing AI risks. With AI development progressing rapidly, experts emphasize that sustained dialogue and action are critical to navigating the future landscape of artificial intelligence.