Criminal Imputation for Violations of the Obligation to Label AI-Generated Content
摘要
人工智能生成内容标识义务违反行为的入罪,具有法益、义务与体系三重依据.侵犯著作权罪、帮助信息网络犯罪活动罪、诈骗罪共犯等传统罪名因主观罪过、行为构造或法益类型错位而均不能适用,拒不履行信息网络安全管理义务罪是唯一具备解释潜力的罪名,但受制于义务归属与结果归属的双重障碍.前者可通过"上位法授权-下位法具体化"的义务链条予以肯定;后者应以隐式标识删除对应"致使刑事案件证据灭失"为主要通道,并以兜底条款对显式标识违反行为作同质性限缩.在此基础上,应按生成、传播、分发三类主体分层设定归责标准,并以责令改正的规范化审查、技术不能与有效合规抗辩作为出罪机制,防止标识义务刑事化异化为对技术创新的过度威慑.
The criminalization of violations of artificial intelligence-generated and synthesized content(AIGC)labeling obligations finds solid justification in legal interests,legal duties,and systemic coherence.Traditional offences—including the crime of assisting information-network criminal activities,joint fraud,and copyright infringement offences—are generally inapplicable because of structural mismatches in mens rea,act types,or protected legal interests.The crime of refusing to perform cybersecurity management obligations is the only offence with genuine interpretive potential,yet its application encounters two obstacles:the attribution of the underlying duty and the attribution of the statutory consequence.The former may be overcome by recognizing a normative chain in which higher-level legislation provides authorization and subordinate rules specify the duty's content.With respect to the latter,the deletion of implicit labels should constitute the principal route to liability under the statutory circumstance of"causing the loss of evidence in a criminal case",whereas explicit-labeling violations should be brought within the residual clause only through restrictive,qualitative homogeneity interpretation.On this basis,attribution standards should be differentiated among generative-service,content-dissemination,and application-distribution entities.Meanwhile,standardized judicial review of corrective orders,technological impossibility,and effective compliance should operate as exculpatory mechanisms,preventing labeling violations' criminalization from exerting an excessive chilling effect on technological innovation.