“深度伪造”违法信息算法传播入罪的困境与破解

Difficulties and solutions to the crime of "deep fake" illegal information algorithm dissemination

成果信息
DC信息
传播影响力
本库下载频次:
本库浏览频次:
CNKI下载频次:0

归属学者：

马方

归属院系：

刑事侦查学院

作者：

王文娟 ; 马方2

摘要：

"深度伪造"违法信息的法律风险在于人工智能算法的深度学习和信息伪造双层次技术结合导致违法信息的算法传播。基于282份网络违法信息传播的裁判文书反映的结果和相关司法解释的分析,可知司法实践中已普遍将传播数量作为入罪标准。随着算法推荐、算法歧视、网络爬虫等人工智能技术的发展,"深度伪造"违法信息算法传播入罪存在诸多司法适用困境。具体而言:单纯"深度伪造"违法信息传播数量并不具有刑法法益侵害性;"深度伪造"传播数量易达至入罪标准,降低入罪门槛;现有技术无法剔除异常传播的数量,计量认定过于宽泛化。为解决"深度伪造"违法信息事实认定方面的司法困扰,本文在规范体系下提出算法传播入罪风险的破解路径。主要包括:"深度伪造"违法信息算法传播行为应具有刑法法益侵害的现实性;"深度伪造"违法信息算法传播数量关联结果与现实危害后果应具有等价性;"深度伪造"违法信息算法传播计量事实的认定应符合相对确凿性,适用优势证据证明标准。

出版日期：

2021-01-10

学科：

刑法学

收录：

北大核心期刊; CSSCI

提交日期：

2021-03-01

引用参考：

王文娟;马方. “深度伪造”违法信息算法传播入罪的困境与破解[J]. 新闻界,2021(01):64-74.

全文附件授权许可：

知识共享许可协议-署名

  • dc.title
  • “深度伪造”违法信息算法传播入罪的困境与破解
  • dc.contributor.author
  • 王文娟;马方
  • dc.contributor.author
  • Wang Wenjuan;Ma Fang;School of Law, Sichuan University;School of Criminal Investigation, Southwest University of Political and Science
  • dc.contributor.affiliation
  • 四川大学法学院;西南政法大学
  • dc.publisher
  • 新闻界
  • dc.publisher
  • Journalism and Mass Communication Monthly
  • dc.identifier.year
  • 2021
  • dc.identifier.issue
  • 01
  • dc.identifier.page
  • 64-74
  • dc.date.issued
  • 2021-01-10
  • dc.subject
  • “深度伪造”;传播数量;违法信息;刑事规制
  • dc.subject
  • "deep fake";amount of transmission;illegal information;criminal regulation
  • dc.description.abstract
  • "深度伪造"违法信息的法律风险在于人工智能算法的深度学习和信息伪造双层次技术结合导致违法信息的算法传播。基于282份网络违法信息传播的裁判文书反映的结果和相关司法解释的分析,可知司法实践中已普遍将传播数量作为入罪标准。随着算法推荐、算法歧视、网络爬虫等人工智能技术的发展,"深度伪造"违法信息算法传播入罪存在诸多司法适用困境。具体而言:单纯"深度伪造"违法信息传播数量并不具有刑法法益侵害性;"深度伪造"传播数量易达至入罪标准,降低入罪门槛;现有技术无法剔除异常传播的数量,计量认定过于宽泛化。为解决"深度伪造"违法信息事实认定方面的司法困扰,本文在规范体系下提出算法传播入罪风险的破解路径。主要包括:"深度伪造"违法信息算法传播行为应具有刑法法益侵害的现实性;"深度伪造"违法信息算法传播数量关联结果与现实危害后果应具有等价性;"深度伪造"违法信息算法传播计量事实的认定应符合相对确凿性,适用优势证据证明标准。
  • dc.description.abstract
  • The legal risk of "deeply faking" illegal information lies in the algorithmic dissemination of illegal information caused by the combination of the deep learning of artificial intelligence algorithms and the two-level technology of information forgery.Based on the results of the 282 online illegal information dissemination judgment documents and the analysis of relevant judicial interpretations,it can be seen that the amount of dissemination has been commonly used as the incriminating standard in judicial practice.The amount of transmission is not all legally intrusive;the quantity of pure "deep forgery" illegal information dissemination does not infringe the legal interests of criminal law;the guilty standard is easily reached based on the amount of transmission as the guilty standard,which reduces the threshold of conviction;the existing technology cannot eliminate the number of distorted transmissions,The measurement recognition is broad.In order to solve the judicial troubles in the identification of "deeply faked" illegal information facts,under the existing normative system,a path for cracking the risk of algorithm dissemination is proposed.Mainly include:the amount of transmission should be intrusive to the legal interests of criminal law;the correlation result of the amount of transmission should be homogenous with the actual harmful consequences;the determination of measurement facts should meet the relative conclusiveness and the application of superior evidence standards.
  • dc.description.sponsorshipPCode
  • 2019CLS2019Y02
  • dc.description.sponsorship
  • 中国法学会2019年度部级法学研究课题“人工智能立法研究”(CLS(2019)Y02)
  • dc.identifier.CN
  • 51-1046/G2
  • dc.identifier.issn
  • 1007-2438
  • dc.identifier.if
  • 0.493
  • dc.subject.discipline
  • D924.3
回到顶部