Will AI safety find a spot at the Trump-Xi talks?

During President Donald Trump's state visit to the United States, China's Xi Jingping is set to meet with him, focusing on potential discussions about artificial intelligence (AI) safety and governance. So far, the intense competition between the US and China has been primarily about creating the most powerful and cost-efficient AI models, as well as gaining control over future advancements in the field.

Recently, the US has proposed a notification mechanism with China for reporting significant AI incidents, with Treasury Secretary Scott Bessent revealing this proposal during talks with Chinese Vice Premier He Lifeng. The primary goal behind this move is to establish a "shared vision of common goals and common threats." The subject of AI risks, such as uncontrollable agents, non-state actors, and cyber non-state actors in bioweapons, could be included in the agenda during their bilateral meeting.

Experts view this as a significant initial step, as some of the emerging risks from advanced AI systems are not confined to national borders. Academic Jean-Marc Rickli of the Geneva Centre for Security explained that such a communication channel could become crucial as AI is utilized in military and security systems.

While the push for more coordination within the AI industry started with government agencies, it has also been driven by industry leaders. Anthropic CEO Dario Amodei called for the industry to "pace the frontier" of AI, highlighting that safety measures are struggling to keep up with AI's rapid advancement. OpenAI CEO Sam Altman, Elon Musk, and Google DeepMind's Demis Hassabis also supported this idea, suggesting that international technical standards for frontier AI should be established to tackle recursive self-improvement and ensure safety.

The differing viewpoints on AI safety among Chinese and US researchers are evident. While Chinese experts have focused more on issues like deepfakes, information security, and social disruption, their counterparts in the US have emphasized the risk of losing control over autonomous AI systems. However, both countries have a shared incentive to discuss AI risks, particularly in mitigating potential security threats.

Despite the strong competition for AI dominance, there is still room for cooperation between the US and China to address risks such as AI-enabled cyberattacks, biological misuse, and major model failures. If the two nations can agree on how to report and communicate during an AI emergency, this could help build technical confidence over time, as suggested by Tsinghua University scholars Sun Chenghao and Xiao Qian.