Turning the Tables: Empowering LLMs to Counter Deceptive Opponents

Marwa Abdulhai*, Jonathan Bodea*, Lennart Luettgau, Christopher Summerfield

*Contributed equally

In Progress (2026)

Summary

We study how LLM agents can detect and defend against deceptive counterparts in strategic dialogue, developing methods that equip models to recognize manipulation and respond robustly when interacting with adversarial opponents.