人工智能算法偏见的法律规制  

Legal Regulation of AI Arithmetic Bias

作　　者：黎平 闫纪阳 LI Ping;YAN Jiyang(Law Department,Jingdezhen Ceramic University,Jingdezhen Jiangxi 333403,China)

机构地区：[1]景德镇陶瓷大学法学系,江西景德镇333403

出　　处：《上饶师范学院学报》2025年第1期25-33,共9页Journal of Shangrao Normal University

基　　金：江西省知识产权保护软科学研究项目(2023-25)。

摘　　要：算法偏见作为人工智能技术深度嵌入社会运行引发的核心伦理与法律问题,其本质是技术系统对社会既有结构性歧视的数字化转译与系统性放大。鉴于算法偏见双重生成机理和算法偏见的多维危害,我国应当构建“预防—救济—惩戒”三位一体的算法偏见法律规制框架:行政监管层面建立分级算法披露制度与动态备案审查机制以平衡商业秘密保护与公众知情权;民事救济层面创新公益诉讼模式,采用过错推定原则,并明确检察机关与网信部门指定组织的诉讼主体地位;刑事规制层面强化算法损害结果的可责性,通过破坏计算机信息系统罪、侵犯公民个人信息罪等罪名惩治恶意算法操纵行为。As the core ethical and legal issue arising from the deep embedding of artificial intelligence technology into social operation,algorithmic bias is essentially the digital translation and systematic amplification of the existing structural discrimination in society by the technological system.Based on the deconstruction of the dual generation mechanism of algorithmic bias and the sorting out of its multi-dimensional harms,in response to the current vacuum of rights relief in the legal regulation of algorithmic bias,China should construct a trinity legal regulation framework of"prevention-relief-punishment":at the administrative supervision level,establish a graded algorithm disclosure system and a dynamic filing review mechanism to balance the protection of trade secrets and the public's right to know;at the civil relief level,innovate the public interest litigation model,adopt the principle of fault presumption and clarify the litigation subject status of the procuratorial organs and the organizations designated by the Cyberspace Administration;at the criminal regulation level,strengthen the culpability of the consequences of algorithmic damage,and punish malicious algorithm manipulation behaviors through charges such as the crime of disrupting computer information systems and the crime of infringing citizens'personal information.

关 键 词：算法偏见 算法披露 公益诉讼 刑事责任 

分 类 号：D912.29[政治法律—民商法学]