I am a tenure-track assistant professor in the Computer Science Department at University of Mississippi. I am the PI of the Reliable Intelligence Systems Lab at University of Mississippi. We are interested in the following questions:
How can we quantify whether an AI system is trustworthy when its goals, environments, and behaviors are uncertain?
How can we prevent attackers from manipulating information to cause incorrect inference?
How can AI systems make safe decisions when evidence is incomplete, ambiguous, or deceptive?
I collaborate closely with Dr. Guangliang Liu at Indiana University and Dr. Zoe Xi Chen at Nanyang Technological University.
- Attack Intent Recognition: Existing identification strategies rely on explicit linguistic cues, but intent itself is an implicit signal. Performing contextual grounding in attack situations and conducting implicit intent inference is a highly challenging problem.
- Safety Alignment: Mechanism Analysis and Improvement: Existing statistical ML methods struggle to achieve intrinsic alignment. I aim to leverage insights from moral psychology to realize intrinsic alignment.
- Pragmatic Reasoning-Based Defense: Language is inherently pragmatic, yet current security research is grounded in semantics, which fundamentally explains why many defense strategies fail to generalize. Our lab is collaborating with researchers in linguistics and NLP to design pragmatic reasoning paradigms for security problems to make large language models safer.
- Security Analysis and Defense of Chatbot Systems: Real-world system security issues are often highly complex. I aim to validate the research across the above directions through attack-and-defense experiments on real systems.
To the best of our knowledge, these are the first generalized moral reasoning models. Huge thanks to all of my collaborators for their contributions and support throughout this project.
Openings: I am looking for self-motivated PhD students for Spring/Fall 2027. Please email me your CV, transcripts, GRE and TOEFL/IELTS scores at bchen5@olemiss.edu.
Updates
- May 2026: I will serve as a program committee member for USENIX Security.
- May 2026: Excited to enjoy the summer together with Dr. Guangliang Liu from Indiana University and Dr. Zoe Xi Chen from Nanyang Technological University on a new computational pragmatics framework for toxic language and malicious attack commands.
- April 2026: Roucheng Ou joined my lab for a research topic about pragmatic gaps in moral alignment.
Lab
Current Members
Roucheng Ou - incoming Ph.D. student
Yawei Li - incoming Ph.D. student
Contact
Email: bchen5@olemiss.edu. Office hours are by appointment.