摘要
以ChatGPT为代表的生成式人工智能引发的虚假有害信息风险极为复杂,涵盖“积极生成型”“消极生成型”和“人为操纵型”等多种情形。依靠其独特复杂的内外部因素,生成式人工智能打破了学术界与实务界迄今关于搜索引擎算法“不能、不想、不愿”生成虚假有害信息的所有立场分歧与理论认知,更是在法律定位、归责认定上对现行网络虚假有害信息相关法律规制发起了挑战。有鉴于生成式人工智能的类通用属性,对其规范应当超出一般算法的规制框架,以人工智能基础性立法为基底建构特定的算法诽谤规制路径。一要坚持层级治理的规制逻辑,建立多元主体协同共管机制、设立涵盖多责任主体的内容监管机制、创建具有高拓展性的风险监测机制;二要在“刺破人工智能面纱”原则的基础上,探索生成式人工智能为责任主体的新型诽谤救济规则,包括探寻以“AI价值链”为导向的归责方法,以及适用于过渡期的暂代型诽谤救济模式。
The risks of disinformation posed by generative AI,exemplified by ChatGPT,are exceedingly intricate,encompassing diverse scenarios such as"active generative type""passive generative type"and"manipulative type."Leveraging its distinctive and complex internal and external factors,generative AI disrupts the prevailing academic and practical consensus and theoretical understanding regarding"the incapacity,disinterest,or reluctance"of search engine algorithms to generate disinformation.Moreover,it presents a formidable challenge to the existing legal framework in the determination and attribution of the responsibility in relation to the regulations of false and harmful information on the internet.Given the quasi-universal attributes of generative AI,its regulation necessitates a framework that transcends the conventional regulatory scope of algorithms,establishing a tailored pathway for regulating algorithmic defamation within the foundational legislation governing artificial intelligence.Firstly,it is imperative to adhere to a hierarchical governance approach,establishing mechanisms for collaborative governance among diverse entities,instituting content supervision mechanisms that encompass multiple responsible parties,and fostering a highly scalable risk monitoring system.Secondly,grounded in the principle of"unveiling the veneer of artificial intelligence,"it is crucial to explore novel defamation remedies that consider generative AI as the accountable entity.This exploration encompasses investigating accountability approaches guided by the"AI value chain"and transitional models of defamation remedies that are applicable during the transitional phase.
出处
《比较法研究》
CSSCI
北大核心
2023年第5期34-54,共21页
Journal of Comparative Law
基金
国家法治与法学理论研究课题“网络侵害法律规制研究”(项目编号:18SFB5006)
江苏省社会科学基金课题“危害物联网安全场域的新型犯罪问题研究”(项目编号:19FXD006)
江苏高校哲学社会科学研究项目“反校园欺凌法律机制研究”(项目编号:2018SJA1295)的阶段性研究成果
作者简介
朱嘉珺,苏州大学王健法学院副教授,法律博士。
