China is advocating for the creation of a multilateral AI cooperation organization to promote human-centric technology, contrasting with the restrictive access models currently pursued by the United States and its allies.

中國正倡導建立一個多邊 AI 合作組織以推動以人為本的技術，與美國及其盟友目前採取的限制訪問模式形成對比。

The strategic orientation of the People's Republic of China (PRC) regarding artificial intelligence is characterized by an emphasis on inclusivity and the support of the Global South. Diplomatic representatives, including Wang Yi and Zhao Haibing, have articulated a preference for open-source, low-cost models to counteract what they characterize as monopolistic and exclusive technological development. This approach is operationalized through the 'AI Capacity Building for All' initiative and the utilization of existing multilateral frameworks, such as the BRICS and the Shanghai Cooperation Organization (SCO), to facilitate technological dissemination.

中華人民共和國（PRC）在人工智慧方面的戰略方向，特點在於強調包容性並支持「全球南方」。包括王毅與趙海兵在內的外交代表表示，他們傾向使用開源、低成本的模型，以對抗其定義為壟斷且排他的技術發展。這種做法是透過「全民 AI 能力建設」倡議，以及利用金磚國家（BRICS）與上海合作組織（SCO）等現有的多邊框架來促進技術傳播。

Conversely, the United States and the Group of Seven (G7) have adopted a more restrictive posture. Reports indicate a deliberative process within the G7 to limit access to advanced AI models to a designated cohort of 'trusted partners.' This systemic divergence is further evidenced by the economic models employed; while U.S. entities predominantly utilize subscription-based access, Chinese initiatives frequently provide downloadable, open-source alternatives.

相反地，美國與七國集團（G7）採取了更為限制的姿態。報告指出，G7 內部正商討將高級 AI 模型的訪問權限限制在特定的「信任夥伴」群體內。這種系統性分歧在採用的經濟模式中 further 得到體現：美國實體主要使用訂閱制訪問，而中國的倡議則經常提供可下載的開源替代方案。

From a geopolitical perspective, the perceived influence of Western narratives embedded within AI training data is viewed by some observers, such as Margarita Simonyan, as a potential threat to the national sovereignty of Russia and China. In response, academic perspectives from China suggest that domestic models, such as DeepSeek, are engineered to operate within the linguistic and cultural parameters of the PRC, thereby ensuring alignment with national values. Despite these ideological frictions, both the U.S. and China have formally acknowledged a mutual intent to establish AI 'guardrails,' although the specific parameters of such an agreement remain undefined.

從地緣政治視角來看，部分觀察員（如 Margarita Simonyan）認為，AI 訓練數據中嵌入的西方敘事對俄羅斯與中國的國家主權構成潛在威脅。對此，中國學界認為，如 DeepSeek 等國內模型是根據中華人民共和國的語言與文化參數而設計，從而確保與國家價值觀一致。儘管存在這些意識形態摩擦，美國與中國均正式承認雙方有意建立 AI 「護欄」，儘管該協議的具體參數仍未定義。

Current global AI governance is defined by a tension between China's push for open, multilateral cooperation and the U.S.-led strategy of controlled, partner-based access.

目前全球 AI 治理的定義，在於中國推動的開放多邊合作與美國主導的受控夥伴訪問策略之間的緊張關係。