摘要： 人工智能（AI）国际治理已形成多主体、多层次与多形式共存的规则体系。在规则数量增加的同时，AI国际治理的规则体系却呈现出结构性失衡。在主体关系层面，规则权威、技术能力与公共责任在不同主体之间发生错配；在治理层次层面，全球规则、区域规则、国内法、行业标准与企业规则之间存在制度竞争，缺乏衔接机制；在治理工具层面，硬法、软法、技术标准、监管沙盒与争端解决机制尚未形成功能闭环。规则体系失衡源于规则权力结构与责任结构的分离、制度竞争对规则公共性的侵蚀以及AI技术逻辑与传统法律逻辑之间的张力。优化规则体系需要通过主体关系的再平衡、治理层次的再贯通与治理工具的再组合，推动AI国际治理从碎片化竞争走向体系化共治。

Abstract: The international governance of artificial intelligence (AI) has formed a rule system characterized by multi-actor, multi-level, and multi-form coexistence. While the number of rules has increased, the rule system of AI international governance exhibits structural imbalance. At the actor relation level, rule authority, technological capacity, and public responsibility are mismatched among different actors. At the governance level dimension, institutional competition exists among global rules, regional rules, domestic laws, industry standards, and corporate rules, with a lack of connecting mechanisms. At the governance tool dimension, hard law, soft law, technical standards, regulatory sandboxes, and dispute settlement mechanisms have yet to form a functionally closed loop. The imbalance of the rule system stems from the separation of rule power structures and accountability structures, the erosion of the publicness of rules by institutional competition, and the tension between the technological logic of AI and traditional legal logic. Optimizing the rule system requires rebalancing actor relations, reconnecting governance levels, and recombining governance tools, so as to shift AI international governance from fragmented competition to systematic co-governance.