Hugo Lyons Keenan

I'm a PhD candidate at the University of Melbourne, working on AI safety and monitoring. My research primarily focuses on understanding AI models better and using this understanding to detect when they are behaving unusually.

Google Scholar GitHub Twitter Email CV

Publications

2026

LLMs Learn to Evade Latent Monitors from Prior Feedback Alone

Hugo Lyons Keenan, Christopher Leckie, Sarah Erfani

NeurIPS 2026 Workshop on Agents in the Wild: Safety, Security, and Beyond (AIWILD)

pdf arXiv code models project page
2026

Mechanistic Anomaly Detection via Functional Attribution

Hugo Lyons Keenan, Christopher Leckie, Sarah Erfani

International Conference on Machine Learning (ICML)

pdf arXiv poster code thread
2025

HALO: Robust Out-of-Distribution Detection via Joint Optimisation

Hugo Lyons Keenan, Sarah Erfani, Christopher Leckie

IEEE Conference on Secure and Trustworthy Machine Learning (SaTML)

pdf arXiv code
2025

A Microsimulation Model of Behaviour Change Calibrated to Reversal Learning Data

Roben Delos Reyes, Hugo Lyons Keenan, Cameron Zachreson

Journal of Artificial Societies and Social Simulation (JASSS)

paper arXiv