生成式人工智能对涉密信息系统安全的促进与挑战

Opportunities and Challenges of Generative Artificial Intelligence for Confidential Information System Security

涉密信息系统是存储、处理国家秘密的计算机信息系统。近年来,随着生成式人工智能的快速发展,其通过已有知识快速推断生成大量数据的“涌现”能力被应用在网络攻防的各个领域,促进了网络安全技术的加速迭代。从涉密信息系统安全要点出发,结合生成式人工智能的现有能力和应用场景,系统地阐述了以大语言模型为代表的生成式人工智能技术给涉密信息系统安全带来的促进作用和挑战。此外,从实践角度出发,针对涉密信息系统管理人员,从生成式人工智能的使用和防护2个方面给出了针对性建议。

Confidential information system refers to the network system that stores,processes and transmits state secret information.In recent years,GAI(Generative Artificial Intelligence)has gained fame,and its ability to quickly infer and generate large amounts of data through existing knowledge leads to its rapid adoption in various fields of network attack and defense,which accelerates the iteration of cyber security technologies.Starting from the key points of security of confidential information systems,combined with the existing capabilities and application scenarios of GAI,this paper systematically expounds the opportunities and challenges brought by GAI represented by big language model to confidential information system security.