风险预防原则在生成式人工智能风险治理中的扩展适用  

The Extended Application of Risk Prevention Principle in Generative Artificial Intelligence Risk Governance

作　　者：刘玉绰 Liu Yuchuo(Law School,Anyang Normal University,Anyang 455000,Henan,China)

机构地区：[1]安阳师范学院法学院

出　　处：《科学.经济.社会》2024年第6期110-124,共15页Science Economy Society

基　　金：河南省高校人文社会科学研究一般项目“数字权力的预防型法治化研究”(2025-ZDJH-020);国家社会科学基金重大研究专项项目“社会主义核心价值观与文化法律制度的构建研究”(19VHJ018)。

摘　　要：生成式人工智能在为人类带来一系列技术红利的同时也为社会埋下了诸多风险隐患,该类潜在风险倒逼风险预防原则在生成式人工智能风险治理中扩展适用。风险预防原则以“风险”和“预防措施”为根本要素,并在两要素之间的逻辑关联下派生出合比例性原则和程序正当性原则两项子原则。在两项子原则的映射下,风险预防原则可推衍出行政权的限制性扩张、风险预防措施的类型化、预防参与主体和规制机制的多元化等内涵。风险预防原则在生成式人工智能风险治理中的扩展适用具有现实基础、制度基础以及其他领域经验基础。为此,在推衍出风险预防原则的内涵以及证成对其扩展适用的必要性基础之上,我们可从观念秉持、主体参与、分类分级管控、成本分析、机制保障、司法保障等维度对风险预防原则在生成式人工智能风险治理中扩展适用予以具化。Generative artificial intelligence not only brings a series of technological dividends to humanity,but also buries many risks and hidden dangers for society.This potential risk forces the Risk prevention principle to be extended and applied in risk management of generative artificial intelligence.The Risk prevention principle takes“risk”and“preventive measures”as the fundamental elements,and derives two sub principles:the principle of proportionality and the principle of procedural legitimacy from the logical relationship between the two elements.Under the mapping of two sub principles,the Risk prevention principle can be derived from the restrictive expansion of administrative power,the typification of risk prevention measures,and the diversification of prevention participants and regulatory mechanisms.The extended application of the Risk prevention principle in generative artificial intelligence risk governance has practical,institutional,and other field experience foundations.Therefore,based on the derivation of the connotation of the principle of risk prevention and the necessity of its extended application,we can concretize the extended application of the Risk prevention principle in generative human intelligence risk governance from the dimensions of concept adherence,subject participation,classification and hierarchical control,cost analysis,mechanism guarantee,judicial guarantee,etc.

关 键 词：风险预防原则 生成式人工智能 风险治理 扩展适用 

分 类 号：D911.04[政治法律—宪法学与行政法学]