生成式人工智能风险分析及治理措施  

Potential Risks Associated with Generative Artificial Intelligence and Governance Measures

作　　者：刘鹏展 王华朋[1] Liu Pengzhan;Wang Huapeng(Criminal Investigation Police University of China,Shenyang 110854,China)

机构地区：[1]中国刑事警察学院,辽宁沈阳110854

出　　处：《贵州警察学院学报》2025年第2期89-97,共9页Journal of Guizhou Police College

摘　　要：Open AI推出生成视频类智能工具Sora,引发了人们对Sora、Chat GPT等为代表的生成式人工智能潜在风险的又一轮讨论浪潮。为探索解决生成式人工智能潜在风险问题的综合性治理措施,从生成式人工智能可能引发的社会风险与犯罪风险两个方面,多层次、多角度对生成式人工智能在就业领域、知识产权保护领域和虚假信息传播领域等社会层面,以及以新型电信网络诈骗、新型智能合约诈骗为例的国家安全层面进行分析。提出生成式人工智能所引起的潜在风险具有复杂性与隐蔽性共存、全新性和未知性叠加、危害性强且不易防范的复杂特征,对国家治理体系和治理能力现代化产生巨大挑战。通过宏观和微观角度对法律法规的完善方向提出建议,确保生成式人工智能技术的健康、可持续发展。The recent release of Sora,OpenAI's video-generating tool,has reignited discussions concerning the potential risks associated with generative artificial intelligence(GAI),exemplified by Sora,ChatGPT and similar AI tools.To explore comprehensive governance measures for potential risks associated with GAI,this study,from the perspectives of social and criminal risks that may be brought about by GAI,analyzes risks in social aspects such as employment,intellectual property protection,and false information dissemination,as well as emerging risks in the national security aspects,including novel telecom network fraud and smart contract fraud enabled by GAI.Then,it identifies three complicated features of GAI-induced risks:coexisting complexity and concealment,novelty and unpredictability,and highly detrimental impacts hard to prevent.These pose significant challenges to the modernization of national governance systems and capabilities,so the paper proposes macro-level and micro-level recommendations for refining legal and regulatory frameworks to ensure the healthy and sustainable development of GAI technologies.

关 键 词：SORA 生成式人工智能 风险治理 

分 类 号：D922.17[政治法律—宪法学与行政法学] TP18[政治法律—法学]