Mujtaba, Dena, Brian Hu, Anthony Hoogs, and Arslan Basharat. 2026. “Aligning Machiavellian Agents: Behavior Steering via Test-Time Policy Shaping”. Proceedings of the AAAI Conference on Artificial Intelligence 40 (44):37738-46. https://doi.org/10.1609/aaai.v40i44.41109.