生成式人工智能内容标识制度的审视与优化
Review and Optimization of Generative AI Content Labeling System
马子斌1
作者信息
摘要
生成式人工智能技术是重要新质生产力,同时也造成机器生成与人类创作内容之间混淆和误认的现实风险,生成式人工智能内容标识制度成为预防风险的靶向路径.现行内容标识制度存在法律规定缺位、履行主体错位、边界限度越位的现实漏洞,树立"生成式人工智能与人类创作协同发展"的本质目的是弥补漏洞的前提.同时,应以"现有技术水平"和"比例原则"为基本原则,在合理限度之内重新建构分类主体下的内容标识义务框架.遵循内容标识的技术机理,合理优化标识规则:通过内容凭证系统,实行"技术审核+人工审核+用户投诉举报"为核心的标识方式,采用"生成式人工智能贡献度+风险文字提示"的标识信息,并参照《商标法》对混淆或误认的标识条件进行合理解释.
Abstract
Generative AI technology is an important new productive force,but it also causes the real risk of confusion and misidentification between machine-generated and human-created content,and the generative AI content identification system has become a targeted path to prevent risks.The current content identification system has practical loopholes such as the lack of legal provisions,the dislocation of the performance subject,and the overstepping of boundary limits,and the essential purpose of establishing"the coordinated development of generative AI and human creation"is the premise of making up for the loopholes.At the same time,the content identification obligation framework under the classification subject should be reconstructed within reasonable limits based on the basic principles of"state of the art"and"principle of proportionality".Follow the technical mechanism of content identification and reasonably optimize the identification rules:through the content voucher system,the identification method with"technical review+manual review+user complaint and report"as the core is implemented,and the identification information of"generative artificial intelligence contribution+risk text prompt"is adopted,and the confusion or misidentification of the marking conditions is reasonably explained with reference to the Trademark Law.
关键词
生成式人工智能/内容标识主体/内容标识义务/内容标识规则Key words
Generative AI/Content Identification Subjects/Content Identification Obligations/Content Identification Rules引用本文复制引用
出版年
2024
NETL
NSTL
万方数据