Security for AI system

AI is increasingly integrated into software systems, development workflows, and user-facing applications. This project studies the security challenges introduced by AI-powered systems, including the reliability, trustworthiness, and attack surfaces of intelligent agents and AI-assisted software.

Our research aims to understand how AI systems interact with real software environments, how adversaries may manipulate these interactions, and how to build analysis and defense techniques for safer AI-driven applications.

People

Yi Qian

2021-now

Xingbang He

2024-now