Security and Resilience Challenges in AI-Centric Systems

Naorin Hossain; Karthik Swaminathan; Pradip Bose

ISCA 2025

Tutorial

21 Jun 2025

Security and Resilience Challenges in AI-Centric Systems

Abstract

In this talk, we discuss several security and resilience challenges surrounding AI-Centric Systems and how these challenges have been addressed by prior works.

Paper

Defensive Prompt Patch: A Robust and Generalizable Defense of Large Language Models against Jailbreak Attacks

Chen Xiong, Xiangyu Qi, et al.

ACL 2025

Workshop paper

Understanding the Error Sensitivity of Privacy-Aware Computing

Matías Mazzanti, Esteban Mocskos, et al.

ISCA 2025

Conference paper

Be Your Own Neighborhood: Detecting Adversarial Examples by the Neighborhood Relations Built on Self-Supervised Learning

Zhiyuan He, Yijun Yang, et al.

ICML 2024

Conference paper

Towards an Open Format for Scalable System Telemetry

Teryl Taylor, Frederico Araujo, et al.

Big Data 2020

View all publications

Abstract

Related

Defensive Prompt Patch: A Robust and Generalizable Defense of Large Language Models against Jailbreak Attacks

Understanding the Error Sensitivity of Privacy-Aware Computing

Be Your Own Neighborhood: Detecting Adversarial Examples by the Neighborhood Relations Built on Self-Supervised Learning

Towards an Open Format for Scalable System Telemetry