摘要
随着人工智能大模型技术的飞速发展,网络安全防御体系的核心架构正经历一场深刻变革。传统防御模式面临着攻防速度严重失衡、人力投人过高,以及检测率与误报率难以兼顾等长期难题。基于自然语言处理、多模态推理及智能体协同能力,大模型在威胁理解、攻击推演等关键环节实现突破,推动防御体系从“规则驱动的被动响应”向“认知驱动的主动决策”加速演进。为此系统分析了大模型在数据安全、安全运营、邮件安全等八大关键领域中的应用场景,梳理其推动网络安全智能化演进的路径,并对未来产业格局的潜在重塑进行展望。
The rapid advancement of large language models(LLMs)is fundamentally reshaping the architecture of cybersecurity defense systems.Traditional approaches are hindered by critical challenges,including the imbalance between attack and defense speed,high operational labor costs,and the ongoing trade-off between detection accuracy and false positives.Leveraging capabilities in natural language processing,multimodal reasoning,and agent-based orchestration,LLMs enable semantic-level threat understanding and dynamic adversarial simulations,transitioning cybersecurity paradigms from rule-based passive responses to cognition-driven proactive defense.This paper explores practical applications of LLMs in eight major security domains-including data protection,security operations,and email security-and outlines their role in the intelligent evolution of cybersecurity systems,while envisioning future transformations across the industry landscape.
作者
刘孟奇
高雅婷
谭晓生
袁洲
刘安
张亚昊
LIU Mengqi;GAO Yating;TAN Xiaosheng;YUAN Zhou;LIU An;ZHANG Yahao(State Grid Information and Telecommunication Branch;Beijing Genius Cyber Tech Co.,Ltd.)
出处
《计算》
2025年第3期31-39,共9页
Computing Magazine of the CCF
基金
国家电网有限公司信息通信分公司软科学课题。
关键词
人工智能
大模型
网络空间安全
威胁检测
安全运营
数据安全
漏洞挖掘
artificial intelligence
large language models
cyber security
threat detection
cybersecurity operation
data security
vulnerability discovery
