生成式人工智能的规制挑战与体系应对  

Regulatory Challenges and Systemic Responses of Generative Artificial Intelligence

作　　者：邱遥堃 Qiu Yaokun

机构地区：[1]上海交通大学凯原法学院 [2]中国法与社会研究院

出　　处：《南大法学》2025年第2期131-145,共15页NanJing University Law Journal

基　　金：2023年国家社会科学基金青年项目“功能主义视角下网络平台算法的法律规制研究”(23CFX031)资助。

摘　　要：生成式人工智能虽然具有强大到不逊于甚至优越于人类智能的内容生产能力,但并未提出本质上不同于以往的规制挑战,仍然可用现行人工智能规制体系来应对,进而促使这一规制体系更加合理与完善:在运行过程方面,生成式人工智能的技术复杂、利益冲突与人机交互进一步加剧了过程规制的内在困难,但应当一方面将过程规制合理定位为辅助性规制措施,另一方面以信任为目标,进行适当披露,并增加用户参与;在输出结果方面,生成式人工智能发挥的功能类似搜索算法,并服务于平台利益,应当继续坚持结果规制的主导地位,要求其安全可控、公平公正,坚持人工智能应以人为本、服务于社会公共利益;在输入数据方面,生成式人工智能更加凸显了个人信息保护、著作权保护、正当竞争秩序维护中数据保护与利用之平衡的重要性,应当在数据合规同时,通过法律解释为模型发展提供更多可用数据,但应当确保相关数据所创造的更大社会效益普遍地惠及更广大的社会群体。Although generative artificial intelligence has strong content production capabilities that are not inferior to or even superior to human intelligence,it has not posed fundamentally different regulatory challenges from the past.The current regulatory system for artificial intelligence can still be used upon it,thereby promoting a more reasonable and perfect regulatory system.In terms of operation,the technical complexity,conflicts of interest,and human-computer interaction of generative artificial intelligence further exacerbate the inherent difficulties of process regulation.However,on the one hand,process regulation should be positioned as an auxiliary regulatory measure,and on the other hand,with trust as the goal,appropriate disclosure should be made and user participation should be increased.In terms of output,the function of generative artificial intelligence is similar to that of search algorithms and serves the interests of the platform.Therefore,it is necessary to continue to adhere to the dominant position of result regulation,requiring it to be safe,controllable,fair,just,people-oriented and serve the public interests of society.In terms of input,generative artificial intelligence has highlighted the importance of balancing data protection and utilization in personal information protection,copyright protection,and maintaining a fair competition order.More available data for model development through legal interpretation while ensuring data compliance should be provided.However,it should be ensured that the greater social benefits created by relevant data generally can deliver to a wider social group.

关 键 词：生成式人工智能 法律规制 算法规制 数据合规 

分 类 号：D92[政治法律—法学]