Utilization of Large Language Models for conformity assessment: Chances, Threats, and Mitigations

Abstract

Assessing the conformity of software in measurement instruments is a laborious process and a major bottleneck in the process of developing new devices.Large Language Models have been shown to effectively handle complex tasks and have the ability to surpass humans with regard to speed and accuracy.However, integrating them into the technology stack can bring major security and privacy risks.This position paper performs a threat modeling in this context.By addressing the discovered confidentiality risks the paper draws a way for safely implementing Large Language Models as an essential tool in the process of conformity assessment.

Ähnliche Arbeiten