Time Table

NRI Secure has consistently gone beyond the traditional boundaries of security, taking on challenges in diverse fields such as blockchain, in-vehicle systems, and even the space domain. In this lecture, we will highlight our efforts in one area that has recently attracted particular attention—AI. First, we will introduce our research on conventional AI, including adversarial attacks against image classification and object detection models. For example, we will discuss methods such as the “adversarial T-shirt,” which may appear unusual to humans but can evade detection by AI, as well as how the results of this research were applied in actual projects. Next, we will turn to generative AI (large language models), which has rapidly become a major focus of social interest. We will share our initiatives in areas such as prompt injection and AI red-teaming. Through this lecture, we aim to convey how NRI Secure tackles uncharted technologies and creates new value through the practice of “offense and defense”.

Speakers:
Kenji Kakimoto（柿本　健司）NRI SecureTechnologies, Ltd. Security Engineer, CISSP Research & Development Center, Service Development Promotion Department