nav emailalert searchbtn searchbox tablepage yinyongbenwen piczone journalimg journalInfo journalinfonormal searchdiv searchzone qikanlogo popupnotification paper paperNew
2025, 06, v.37 38-52
从安全防控到发展行动:全球人工智能治理范式的转向
基金项目(Foundation): 国家社科基金重大项目“新时代中国国家安全法治体系和能力现代化研究”(编号:24&ZD120)
邮箱(Email):
DOI: 10.13975/j.cnki.gdxz.2025.06.003
摘要:

从2023年到2025年,三届全球AI峰会的议程,经历了从设置安全防控主题到“安全—发展”二元议题初步整合再到突出强调发展行动的嬗变,在一定程度上揭示出全球人工智能治理从安全防控范式向发展行动范式的转向。作为系统性工程,全球AI治理范式的转向主要体现为目标、主体、行动等系统构成要素的转向,即从安全到发展的目标转向,国家、企业与劳动者的主体转向,从抽象原则到具体规则的行动转向。全球AI治理范式各构成要素发生转向的具体动因不同,谋求数字技术红利的治理需求推动了从安全到发展的目标转向;国家间安全与发展的零和博弈影响治理效能推动了国家身份降维与企业、劳动者身份升维的主体转向;参与全球AI治理的各国际主体政策的分化则在一定程度上推动了从抽象原则到具体规则的行动转向。在全球AI治理范式的转向进程下,先发国际主体与后发国际主体分别持“独占优势”与“共同发展”的治理话语,先发国际主体构建排他性技术权力架构的意图因独占优势的流失而较难落地,后发国际主体秉持共同发展诉求争取更多发展空间,国际政治经济秩序也在两者的治理话语对峙中进行着隐性调整。

Abstract:

From 2023 to 2025, the agendas of the three consecutive Global AI Summits have undergone a transformation from setting “security prevention and control” as the theme to initially integrating the dual themes of "security-development", and then emphasizing “development action”. This to some extent reveals the shift of global AI governance from a “security prevention and control” paradigm to a “development action” paradigm. As a systematic endeavor, the shift in global AI governance paradigm is mainly reflected in the shift of system components such as goals, subjects, and actions, that is, the shift of goals from security to development, the shift of the subjects of countries, enterprises, and workers, and the shift of action from abstract principles to concrete rules. The specific reasons for the shift in various components of the global AI governance paradigm are different, and the governance demand for seeking digital technology dividends has driven the goal shift from security to development; The zero-sum game of security and development between nations has impacted governance effectiveness, driving a shift in the focus between reducing national identity and upgrading the identities of enterprises and workers; The differentiation of policies among international entities involved in global AI governance has to some extent driven the shift from abstract principles to concrete rules in action. In the process of shifting the global AI governance paradigm, the early-mover and late-mover of international entities hold governance discourses of “exclusive advantage” and “common development” respectively. The intention of the early-mover of international entities to construct an exclusive technological power structure is difficult to implement due to the loss of exclusive advantages, while the later-mover of international entities uphold the common development demands and strive for more development space. The international political and economic order is also undergoing implicit adjustments in the confrontation between the two governance discourses.

参考文献

[1]US and UK Refuse to Sign Paris Summit Declaration on “Inclusive” AI[N].The Guardian,2025-02-11.

[2]Britain Publishes “Bletchley Declaration” on AI Safety[N].Reuters,2023-11-01.

[3]What We Learned from the Global AI Summit in South Korea[N].The Guardian,2024-05-28.

[4]Michael Veale,Frederik Zuiderveen Borgesius.Demystifying the Draft EU Artificial Intelligence Act[J].Computer Law Review International,2021,(4).

[5]Nick Bostrom.Superintelligence:Paths,Dangers,and Strategies[M].Oxford:Oxford University Press,2014.

[6]严驰.生成式AI大模型的全球治理方案:何以可能与何以可为[J].科技与法律(中英文),2024,(1).

[7]Luciano Floridi.The Ethics of Artificial Intelligence:Principles,Challenges,and Opportunities[M].Oxford:Oxford University Press,2023.

[8]Anu Bradford.The Brussels effect:How the European Union Rules the World[M].Oxford:Oxford University Press,2020.

[9]李建国,范策.全球生成式人工智能政策规范的核心议题及其世界影响[J].国际观察,2025,(1).

[10]卓华,王明进.技术地缘政治驱动的欧盟“开放性战略自主”科技政策[J].国际展望,2022,(4).

[11]高奇琦.GPT技术的全球治理方案:一个技术多边主义的框架[J].当代世界,2023,(5).

[12]翟志勇.论数据信托:一种数据治理的新方案[J].东方法学,2021,(4).

[13][德]乌尔里希·贝克.风险社会:新的现代性之路[M].张文杰译,南京:译林出版社,2018.

[14]廖理,戚航,闫竹,等.防范金融风险保护金融创新之一——英国监管沙盒调研与建议[J].清华金融评论,2018,(3).

[15]涂良川,潘依林.技术封建主义:科技进步的政治退步——当代西方左翼“技术封建理性批判”刍议[J].世界社会科学,2023,(2).

[16]杨伟,刘健,武健.“种群-流量”组态对核心企业绩效的影响——人工智能数字创新生态系统的实证研究[J].科学学研究,2020,(11).

[17]Perrigo B.Exclusive:OpenAI Used Kenyan Workers on Less than 2 per Hour to Make ChatGPT Less Toxic[EB/OL].https://time.com/6247678/openai-chatgpt-kenya-workers/,2025-10-29.

[18]赵觉珵,任重.美军动向:生成式AI军事化[N].环球时报,2024-01-30.

[19]杨忠,巫强,宋孟璐,等.美国《芯片与科学法案》对我国半导体产业发展的影响及对策研究:基于创新链理论的视角[J].南开管理评论,2023,(1).

[20]石健,李建慧.美国半导体产业“友岸外包”策略研究[J].中国集成电路,2023,(8).

[21]钮维敢,陈康.冷战遗留视角下的美国对华科技遏制战略研究[J].国际观察,2023,(3).

[22]Google's Android Antitrust Appeal Faces Major Blow at Europe's Top Court[EB/OL].https://www.grcreport.com/post/googles-android-antitrust-appeal-faces-major-blow-at-europes-top-court,2025-10-29.

[23]Another Monopoly in Making?OpenAI's Aggressive Tactics Raise Concerns[EB/OL].https://www.tice.news/enticing-angle/another-monopoly-in-making-openais-aggressive-tactics-raise-concerns-7239562,2025-10-29.

[24]王世达,杨冰夷.印度以人工智能推进工业4.0进程[J].信息通信技术与政策,2025,(1).

[25]王珩.非洲人工智能发展态势与中非合作[J].当代世界,2024,(11).

[26]金砖国家领导人第十六次会晤喀山宣言(摘要)[N].人民日报,2024-10-25.

[27]蔡艺生,翁春露.国家安全法治“域外体系”的理论阐释与理性建构[J].中国人民公安大学学报(社会科学版),2022,(5).

[28]谢琛,潘锐.美国主导的多边出口管制合作——以对华芯片封锁为例[J].美国研究,2023,(6).

[29]Nvidia Sounds Fresh Warning about Damage from China Export Rules[N].Bloomberg,2023-08-24.

[30]Nvidia's Jensen Huang Thinks U.S.Chip Curbs Failed — And He's Not Alone[EB/OL].https://www.cnbc.com/2025/05/22/nvidias-jensen-huang-semiconductor-experts-think-us-chip-curbs-failed.html,2025-10-29.

[31]丁军,王臻,谷棣.美国硅谷为何被五角大楼政治化[N].环球时报,2022-09-29.

[32]Andrew W.Wyckoff.A New Institution for Governing AI?Lessons from GPAI[EB/OL].https://www.brookings.edu/articles/a-new-institution-for-governing-ai-lessons-from-gpai/,2025-10-29.

[33]Implementation of Additional Export Controls:Advanced Computing and Semiconductor Manufacturing Items[N].Federal Register,2022-10-13.

[34]刘鹏.非洲个人数据保护立法稳步推进[N].中国社会科学报,2022-08-19.

[35]积极推动全球人工智能包容普惠发展[N].人民日报,2024-09-27.

[36]Colombia's Digital Tax Policy[EB/OL].https://itif.org/publications/2025/02/11/colombia-digital-tax-policy/,2025-10-29.

[37]武长海,黄静怡.欧盟数据治理模式扩张与中国因应[J].科技与法律(中英文),2024,(6).

[38]王栋,王耀正.美俄对后苏联空间国家制度变化的行动与态度——制度扩散还是地缘政治利益[J].国际展望,2022,(6).

[39]夏磊,黄子非,苏航.GS1系统支撑欧盟产品数字护照的落地[J].条码与信息系统,2024,(6).

[40]Global AI Fund Needed to Help Developing Nations Tap Tech Benefits,UN Says[N].The Guardian,2024-09-19.

[41]Qure.AI Introduces Lung Cancer AI Training[EB/OL].https://www.itnonline.com/content/qureai-introduces-lung-cancer-ai-training,2025-10-29.

① 监管沙盒(Regulatory Sandbox)是一种“安全实验区”模式,允许企业在受控环境中测试创新的AI技术,监管部门通过观察风险调整监管政策。

基本信息:

DOI:10.13975/j.cnki.gdxz.2025.06.003

中图分类号:D815;TP18

引用信息:

[1]翁春露.从安全防控到发展行动:全球人工智能治理范式的转向[J].公共治理研究,2025,37(06):38-52.DOI:10.13975/j.cnki.gdxz.2025.06.003.

基金信息:

国家社科基金重大项目“新时代中国国家安全法治体系和能力现代化研究”(编号:24&ZD120)

发布时间:

2025-10-29

出版时间:

2025-10-29

网络发布时间:

2025-10-29

检 索 高级检索

引用

GB/T 7714-2015 格式引文
MLA格式引文
APA格式引文