生成式人工智能安全风险治理研究

Research on Security Risk Management of Generative Artificial Intelligence

当前,以ChatGPT、Sora等现象级产品为代表的生成式人工智能快速发展,正在重塑产业链和价值链。与此同时,也在数据安全、算法安全、网络安全和应用安全等方面引发了诸多新风险,可能造成严重影响。世界主要国家和地区高度重视生成式人工智能安全风险治理,从政策引导、技术保障、国际合作等维度推进风险治理体系建设。科技企业在积极推进前沿技术创新的同时,也应高度重视安全治理,从不同侧重点发力保障人工智能可信赖、负责任。我国应从加强跟踪研究、推进技术攻关、推动“以评促建”和加强国际交流四个维度发力,保障人工智能安全健康发展。

Currently,the rapid development of generative artificial intelligence,represented by phenomenal products such as ChatGPT and Sora,is reshaping the industrial and value chains.At the same time,it has also triggered numerous new risks in areas such as data security,algorithm security,network security,and application security,which may have serious impacts.Major countries and regions worldwide attach great importance to the management of security risks associated with generative artificial intelligence,promoting the construction of risk management systems from dimensions such as policy guidance,technical support,and international cooperation.While actively promoting frontier technological innovation,technology companies also attach great importance to security management,focusing on different aspects to ensure that artificial intelligence is trustworthy and responsible.China should focus on four dimensions:strengthening tracking research,promoting technological breakthroughs,promoting“evaluation to promote construction,”and enhancing international exchanges,to ensure the safe and healthy development of artificial intelligence.