ChatGPT网络安全风险的形成机制及其应对路径

Formation Mechanism of ChatGPT Cybersecurity Risks and the Path of Response

以ChatGPT为代表的生成式人工智能在带来提高效率、增强用户体验等好处的同时,亦会造成严重的网络安全风险。结合ChatGPT的特点和运行原理,分析ChatGPT网络安全风险形成机制,有助于正确认识人工智能技术迭代升级带来的网络安全风险和提出针对性应对路径。在风险生成逻辑上,ChatGPT网络安全风险集中于“生成—使用—再应用”三个阶段,各阶段生成逻辑并不相同;在风险样态及演变趋势上,存在针对性优化多种网络攻击模式、持续性模糊数据边界形成漏洞、颠覆性改变网络数据利用模式等典型风险样态,呈现资本冲突加剧安全风险、虚拟风险渐化为社会现实风险、政治竞争增加不稳定因素等演变趋势;在风险应对路径上,可从减少算力上下限差距实现平衡发展、实施本地部署与云端审计保障数据安全、利用人工智能技术为网络安全防御助力、依靠国家力量全面调控等方面综合着力应对。

While generative artificial intelligence represented by ChatGPT brings benefits such as improved efficiency and enhanced user experience,it also poses serious cybersecurity risks.Combining the characteristics and operation principles of ChatGPT,the analysis of the formation mechanism of ChatGPT cybersecurity risks helps understand the cybersecurity risks brought by the iterative upgrade of artificial intelligence technology correctly and propose a targeted response path.In terms of risk generation logic,ChatGPT cybersecurity risks are concentrated in three stages,i.e.,generation,use and reapplication,with the generation logic different in each stage.In terms of risk patterns and evolution trends,there are targeted optimisation of multiple network attack modes,continuous blurring of data boundaries to form vulnerabilities,and disruptive changes in network data utilisation patterns;and evolution trends such as capital conflicts exacerbating security risks,virtual risks gradually turning into social reality risks,and political competition increasing instability factors,etc.The risk response path can combine such measures as reducing the upper and lower limits of computing power to achieve balanced development,implementing local deployment and cloud auditing to ensure data security,using artificial intelligence technology to enhance cybersecurity defence,and relying on national power for comprehensive regulation and control.