Home // AISyS 2024, The First International Conference on AI-based Systems and Services // View article

Cybersecurity Concerns of Artificial Intelligence Applications on High-Performance Computing Systems

Authors:
Rishabh Saxena
Aadesh Baskar
Sameer Haroon
Sameed Hayat
Oleksandr Shcherbakov
Kerem Kayabay
Dennis Hoppe

Keywords: High-Performance Computing (HPC), Artificial Intelligence (AI), AI Security Vulnerabilities, TOE Framework

Abstract:
The High-Performance Computing (HPC) landscape is undergoing profound changes with developments in fast-growing domains such as Artificial Intelligence (AI), cloud, edge computing, and quantum computing. The growth of AI has particularly impacted the relatively isolated HPC realm, bringing in new user communities like start-ups that don’t want to fall behind and are increasingly dependent on foundational models trained by a handful of companies. However, the rapidly growing AI technology landscape introduces security vulnerabilities to the HPC world, which hesitates to install and maintain potentially unstable software. This paper is a first step towards enabling secure AI workloads on HPC systems by investigating AI security vulnerabilities using the AI Lifecycle. We then organize the challenges for HPC centres through the lens of the Technology-Organization-Environment (TOE) framework. Lastly, we discuss the differences between AI security concerns and mitigation strategies on HPC and other systems, and outline future work towards secure AI workloads on HPC systems.

Pages: 12 to 21

Copyright: Copyright (c) IARIA, 2024

Publication date: September 29, 2024

Published in: conference

ISBN: 978-1-68558-192-3

Location: Venice, Italy

Dates: from September 29, 2024 to October 3, 2024