Reference:

AI Agent Security & Safety

We are seeking candidates to join our AI security group at Lancaster University, and become part of this rapidly growing research field.

This position focuses on AI Agent Security, and specifically Shadow AI. ShadowAI – the deployment and use of Artificial Intelligence within an organisation without knowledge or approval from security teams – is a major issue within companies, given the rapid uptake of agents within organisations. ShadowAI is often hidden or embedded within software applications, and introduces threats spanning model backdoors, coordinated attacks, AI safety issues, and data leakage. This is particularly problematic given the introduction of ShadowAI at the computing edge to the wider organisation. Current tools and methods proposed across both academia and industry are unable to address this issue.

In this project, you will research how to defend AI agents from cyber attack....