美中两国正将人工智慧视为经济与国安竞争的核心，美国持续严格限制先进AI晶片出口，并著手防范中国透过走私、境外云端租赁算力等途径规避管制；同时，美方亦指责多家中国企业未经许可，利用大规模模型蒸馏技术复制美国顶尖AI模型，引发知识产权与监管机制上的巨大争议。

随著具备自主行动与自我改进能力的AI代理崛起，业界对失控及灾难性风险的担忧日益加剧。Anthropic执行长Dario Amodei等人呼吁加强监管并放缓前沿模型开发，但此观点遭川普斥为骗局，亦被北京批评为散布恐慌；不过，双方在防止AI落入恐怖组织、生化武器制造及核指挥系统误判等防护热线方面，仍存在潜在合作空间。

在硬体应用与军事领域，美中在人形机器人及军事AI方面激烈角力。中国在稀土磁铁与致动器等关键供应链占据主导地位，并反制美国的进口限制；同时，美国加强防堵中国军工相关院校取得辉达晶片，双方缺乏互信与共同安全标准的现状，使这场高科技竞赛的紧张局势持续升温。

The United States and China view artificial intelligence as the centerpiece of their economic and national security rivalry, with Washington enforcing strict export curbs on advanced chips while moving to close loopholes involving smuggling and remote cloud computing access; concurrently, the US has accused Chinese firms of relying on large-scale model distillation to replicate leading American models, sparking friction over IP and oversight.

The emergence of autonomous AI agents capable of self-improvement has heightened global concerns over catastrophic risks, prompting figures like Anthropic CEO Dario Amodei to advocate for stricter government regulation and development pauses, which drew pushback from Donald Trump and accusations of fearmongering from Beijing; nevertheless, mitigating threats like cyberattacks, bioweapons, or automated nuclear escalations remains a rare potential avenue for bilateral dialogue.

In the hardware and defense domains, competition is intensifying over AI-powered humanoid robotics and military applications. While China leverages its dominance in key supply chain components like rare-earth magnets against US import barriers, Washington continues to block Chinese defense-linked universities from acquiring cutting-edge Nvidia chips, leaving the high-stakes AI race fraught with geopolitical friction and a critical deficit of mutual trust.