探讨中美在人工智慧领域的竞争，通常被视为零和博弈。然而，随著人工智慧模型（尤其是AI代理）能力的增强，两国的研究人员正试图在人工智慧安全方面展开合作。这种合作不仅是为了解决日益增加的网路安全威胁，更是为了避免潜在的灾难性后果。

中国的开源模型在成本较低的情况下，逐渐缩小与美国尖端模型的差距。中国在推动人工智慧发展时，也面临著严格的监管，更倾向于开发稳定、可靠且具备经济实用性的模型。这与美国对通用人工智慧（AGI）的无止境追求形成鲜明对比，但双方都对系统失控的风险抱持共同的担忧。

尽管政治局势紧张且美国实施出口管制，两国之间的学术和企业合作仍有其必要性。建立沟通机制和互信，有助于防范金融闪电崩盘或AI骇客攻击等重大系统性问题。此外，面对晶片禁令，中国正积极研发替代硬体，显示出美国单纯依赖制裁并不足以维持领先，更需要投资于自身的基础科学与技术创新。

The competition between the US and China in artificial intelligence is often framed as a zero-sum game. However, as the capabilities of AI models, especially AI agents, continue to increase, researchers from both nations are attempting to collaborate on AI safety. This cooperation is driven not only by the need to address growing cybersecurity threats but also by the necessity of avoiding potential catastrophic consequences.

China's open-source models are gradually closing the gap with cutting-edge US models at a lower cost. While promoting AI development, China also faces strict regulations and leans toward creating stable, reliable, and economically useful models. This contrasts with the relentless American pursuit of artificial general intelligence (AGI), yet both sides share a common apprehension about the risks of out-of-control systems.

Despite political tensions and US export controls, academic and corporate collaboration between the two countries remains necessary. Establishing communication mechanisms and mutual trust can help prevent major systemic issues such as financial flash-crashes or AI-driven hacking sprees. Furthermore, in response to chip bans, China is actively developing alternative hardware, demonstrating that relying solely on sanctions is insufficient for the US to maintain its lead; investing in fundamental science and technological innovation is essential.