Bonjour, Hello, مرحبًا, 你好, Привет 🗺️
I’m Loic, a PhD candidate in Computer Science at Mohamed Bin Zayed University of Artificial Intelligence (MBZUAI) in Abu Dhabi, specializing in AI security with a focus on cognitive security for autonomous agents.
My background spans social and cognitive sciences and eight years as an analyst—including childhood protection work within the justice system. While initially drawn to profiling and psychological analysis, I realized AI researchers attempting to model cognition presented a new frontier—one where my skills could be applied to analyzing the systems themselves. This led me to mathematics, the language underlying AI cognition, where I discovered profound connections to psychology and cognitive science. After completing an MSc in AI Engineering, the emergence of autonomous agents convinced me that deeper research was essential.
Inspired by researchers like Nicholas Carlini, I found my calling in AI security, particularly in understanding the cognitive layer of AI systems. My research examines the risks in agent reasoning, belief formation, world modeling, and decision-making—from emergent misalignment and deception to adversarial manipulation of these fundamental cognitive processes.
Latest Posts
Modeling Target Behavior: How World Models Improve Agentic Red-Teaming