引用本文
  • 王志,尹捷,崔翔,刘奇旭,刘潮歌,汪旭童.人工智能赋能网络威胁研究进展[J].信息安全学报,已采用    [点击复制]
  • Wang Zhi,Yin Jie,Cui Xiang,Liu Qixu,Liu Chaoge,Wang Xutong.AI-Powered Cyber Threats: A Survey[J].Journal of Cyber Security,Accept   [点击复制]
【打印本页】 【在线阅读全文】【下载PDF全文】 查看/发表评论下载PDF阅读器关闭

过刊浏览    高级检索

本文已被:浏览 446次   下载 0  
人工智能赋能网络威胁研究进展
王志1, 尹捷1, 崔翔2, 刘奇旭1, 刘潮歌1, 汪旭童1
0
(1.中国科学院信息工程研究所;2.广州大学网络空间先进技术研究院)
摘要:
近年来,人工智能赋能的网络威胁日趋增多。为了理解这类威胁的原理、研究其防御方法,本文对现有的人工智能赋能的网络威胁案例进行了梳理和综述。本文首先分析了人工智能的能力和神经网络模型的性质,总结了人工智能对网络威胁的五种赋能作用,包括伪造与欺骗、隐蔽与隐匿、感知与决策、定向与定制、规模化与自动化,并在此基础上形成了人工智能对网络威胁的赋能矩阵。随后,本文将现有的人工智能赋能的网络威胁案例归纳为18个类别,并结合网络杀伤链模型构建智能化网络威胁框架,从网络攻击的准备、入侵和执行三个阶段对相关案例进行介绍和分析。随后,本文从攻防能力角度分析现有防御方法的有效性与局限性,指出智能化网络威胁与其他威胁的不同,并从场景、技术和系统三个维度提出针对性防御建议。最后,本文结合人工智能领域与网络威胁领域的技术方向,对智能化网络威胁的发展趋势进行展望,对人工智能在网络威胁中的作用进行深入探讨,以期对未来相关防御研究提供有益参考。
关键词:  人工智能 网络威胁 人工智能安全 网络空间安全
DOI:
投稿时间:2022-04-12修订日期:2022-06-15
基金项目:国家自然科学基金项目(面上项目,重点项目,重大项目)
AI-Powered Cyber Threats: A Survey
Wang Zhi1, Yin Jie1, Cui Xiang2, Liu Qixu1, Liu Chaoge1, Wang Xutong1
(1.Institute of Information Engineering, Chinese Academy of Sciences;2.Cyberspace Institute of Advanced Technology, Guangzhou University)
Abstract:
In recent years, the threat of artificial intelligence (AI) in cyber attacks has been continuously increasing. In order to help researchers quickly understand the relevant principles and conduct research on defense methods, it is necessary to analyze the characteristics of AI and the principles of AI-powered cyber threats and sort out relevant cases. To this end, we analyzed the capabilities of artificial intelligence and the nature of the neural network model and divided the roles of artificial intelligence in cyber threats into five categories: forgery and deception, stealth and anonymity, perception and decision-making, targeting hand customization, and scale and automation. On this basis, the influence of the characteristics of artificial intelligence on the five roles is analyzed, and the enabling matrix of artificial intelligence to cyber threats is formed. Then, we collected the existing AI-powered cyber threat works and classified the cases into 18 categories. Combined with the cyber kill chain, an AI-powered cyber threat framework is formed, and the cases of AI-powered cyber threats are introduced based on the three stages of an attack: preparation, intrusion, and execution. The principles, effects, and progress of representative works in each category are included, as well as their strength and limitation. Subsequently, we analyzed the effectiveness and limitations of existing defense methods from the perspective of the capabilities of attackers and defenders, pointed out the differences between AI-powered cyber threats and other threats, and put forward the possible defense measures targeted at AI from three dimensions of scenario, technology, and system. Combined with the evolution of AI technology and cyber threats and the deficiencies and constraints of AI-powered cyber threats, we discussed the effectiveness of AI in cyber threats and prospected the future trends of AI-powered cyber threats. We hope this paper will help defend against AI-powered cyber threats in the future.
Key words:  artificial intelligence  cyber threat  AI security  cyber security