Zero-Day Vulnerabilities in Container Images: Risks and Detection using Generative AI

Karthikeyan Thirumalaisamy

Abstract: Containerization has changed the way we deploy software by giving us the ability to deploy lightweight, portable, and scalable applications. However, it has also introduced a larger attack surface were risk surfaces from usage of unpatched or unknown vulnerabilities in containers. Zero-day vulnerabilities are those flaws that are exploited before being made public or patched, they can pose a serious threat to anything running in a containerized environment because they can easily propagate without any detection. Conventional security scanners are heavily reliant on the known vulnerability databases and do not detect zero-day risks. This paper examines the zero-day vulnerabilities, describes the boundaries of known detection mechanisms, and proposes a new way to identify and reduce risks utilizing generative AI. Using large language models (LLMs) and generative systems, it illustrates how generative AI can enhance both static and dynamic analysis, automate threat pattern recognition, identify relationships between threats, and produce real-time contingent remediation. We argue that generative AI as a solution in DevOps provides a better level of proactivity and reactivity to address the ever-evolving threat landscape in containerized applications.

Keywords: Zero-day, Zero-day vulnerabilities, Container Image vulnerabilities, AI mitigation, AI detection

How to Cite?: Karthikeyan Thirumalaisamy, "Zero-Day Vulnerabilities in Container Images: Risks and Detection using Generative AI", Volume 14 Issue 6, June 2025, International Journal of Science and Research (IJSR), Pages: 309-317, https://www.ijsr.net/getabstract.php?paperid=SR25603123105, DOI: https://dx.doi.org/10.21275/SR25603123105