通用人工智能提供者内容审查注意义务的证成  

Justification of General Artificial Intelligence Providers'Duty of Care in Content Review

作　　者：高阳[1] Gao Yang

机构地区：[1]上海对外经贸大学

出　　处：《东方法学》2024年第1期189-200,共12页Oriental Law

基　　金：国家社会科学基金青年项目“企业数据合理使用的边界及其制度构建研究”(项目批准号:22CFX028)的阶段性研究成果。

摘　　要：通用人工智能可自发或受用户操控被动生成有害信息,它的提供者作为新型的网络内容提供者,负有数字内容“守门人”职责,同时负有网络空间的安全保护义务,应承担内容审查的注意义务,从而预防有害信息的侵权风险。“以数据为中心”的内容生产方式和算法黑箱效应降低了通用人工智能提供者对内容的管控力,导致对生成内容的普遍审查难以被实现。应针对有害信息,结合内容生产场景、提供者风险预见能力差异化构建内容审查的注意义务。该注意义务以技术性审查为标准,提供者采取符合技术特点的内容审查机制,便应当认定履行了相关义务。Artificial General Intelligence can spontaneously or passively generate harmful information under user control.As a new type of content provider,its provider has to be the"gatekeeper"of digital contents.It also has the obligation to protect the security of cyberspace and should have the duty of care in content review to prevent the infringement risks in harmful information.The"data-centered"content production means and algorithmic black box reduce the ability of Artificial General Intelligence providers to control the contents,making it difficult to achieve universal review of generated contents.The duty of care for content review should be differentiated based on harmful information,combined with content production scenarios and the provider's risk foresight capabilities.This duty of care is based on technical review.If the provider adopts a content review mechanism that meets the technical characteristics,it should be deemed that it has fulfilled the relevant obligations.

关 键 词：通用人工智能 有害信息 注意义务 信息内容安全 网络内容提供者 内容审查义务 

分 类 号：DF41[政治法律—经济法学]