| 英文摘要 |
Current international governance of artificial intelligence primarily encompasses three objective dimensions: product liability, ethical values, and national security. The European Union, China, and the United States respectively represent three typical governance pathways—one centered on individual rights, one balancing multiple objectives, and one dominated by national security. These divergent approaches lead to significant differences in the hierarchy of goals and the institutional emphases across the three dimensions. Global AI governance should take these three objectives as its core, building a layered and coordinated governance system through institutional alignment and regulatory dialogue. Specifically, it should begin from a foundation of shared ethical principles, focusing first on product responsibility and system safety, and seek to establish a basic consensus through technical standards, industry norms, and compliance guidelines. On this basis, it should progressively promote the development of regional ethical agreements and soft-law instruments that concretize rights and obligations, thereby fostering higher-level value convergence and institutional cooperation. In matters involving national security, it is essential—against the backdrop of increasingly dual-use (civil-military) AI applications—to explore the creation of transnational crisis management and risk communication mechanisms, so as to prevent excessive militarization or technological containment from impeding global consensus. In this way, the international governance of artificial intelligence can evolve toward an order that takes security as the baseline and cooperation as the guiding principle. |