涉生成式人工智能數據犯罪刑法規制新路徑

A New Path of Criminal Law Regulation Concerning Generative Artificial Intelligence Data Crime

我國數據犯罪刑法規制的側重點在於數據控制行為的規制以及數據分類分級的保護。生成式人工智能的應用導致數據控制行為合法性增強、數據利用行為風險攀升以及數據共享需求提升。生成式人工智能的應用可能會引發現行數據分類分級標準的功能性失靈並造成刑法在法益保護上的漏洞。應當確立“數據利用”行為規制觀與“全類別＋分類分級”數據治理觀。現行數據犯罪刑法規制模式屬於權利保護模式，應當採用“權利保護＋集體法益保護”的複合模式。複合模式保護的集體法益應當是數據管理秩序而非數據安全。應當修改拒不履行信息網絡安全管理義務罪的構成要件以應對生成式人工智能“自發”侵害數據法益的情形。應當增設非法分析數據罪、操縱數據罪以應對行為人利用生成式人工智能侵害數據法益的情形。

Data-based digital technology has become an important technical tool for the rapid development of politics, economy and society in the era of digital economy. In order to protect data in time, China's data crime charge system has been initially formed. At this stage, generative artificial intelligence has revolutionized the way of obtaining and utilizing data in the past, and the criminal regulation of data crime based on the current data crime charge system may not be able to achieve the expected effect. From the relevant provisions of the current criminal law and its advanced law, the emphasis of the criminal regulation of data crime lies in the regulation of data control behavior and the protection of data classification and grading. However, in the application of generative artificial intelligence, the data obtained by generative artificial intelligence is basically open data or self-analyzed data and does not involve data tampering and destruction, resulting in enhanced legitimacy of data control behavior. The scale and clustering of data utilized by generative artificial intelligence have also led to a sharp rise in the behavioral risk of data utilization. In addition, with the rapid development of the digital economy, the demand for data sharing has also increased. Therefore, criminal law focusing on the regulation of data control behavior can't meet the current data legal interest protection and data use needs and should establish the ''data use'' behavior regulation view. It should also be noted that the application of generative artificial intelligence may lead to the functional failure of the current data classification and cause legal interest protection loopholes, and a full category + classification and grading'' data governance concept should be established. From the perspective of the criminal law regulation mode of data crime, the current regulation mode adopted by the criminal law cannot realize the extensive protection of data in the application of generative artificial intelligence and will ignore the regulation of data utilization behavior, so the compound mode of ''right protection + collective legal interest protection'' should be adopted. In order to avoid unreasonable restrictions on the development and application of emerging technologies such as generative artificial intelligence, the collective legal interest of composite model protection should be data management order rather than data security. The constitutive elements of the crime of refusing to perform the information network security management obligation should be modified to deal with the situation of generating artificial intelligence ''spontaneously'' infringing on the legal interests of data, and the situation of refusing to perform the information network security management obligation in Article 286, Item 2 of the Criminal Law should be modified to ''causing the disclosure of classified information, resulting in serious consequences''. Moreover, add ''illegal access to classified information, data'' ''illegal analysis of information, data'' ''tampering with information, data'' ''manipulation of information, data'' four cases. In addition, the crime of illegal data analysis and data manipulation should be added to deal with the situation where the actor infringes on the legal interests of data by using generative artificial intelligence. When judging the ''serious circumstances'' in the crime of illegal data analysis and data manipulation, it may be judged by the number and frequency of illegal analysis and data manipulation by the perpetrator, as well as whether it involves state secrets, trade secrets, securities or futures trading and other information sensitive fields.