一作ACL 2026Poster· Natural Language Processing
Multimodal Safety Evaluation in Generative Agent Social Simulations
Alhim Adonai Vera Gonzalez · Carlos Hinojosa · Karen Sanchez · Haidar Bin Hamid · Donghoon Kim · Bernard Ghanem
摘要
Can generative agents be trusted in multimodal environments? Despite recent advances, agents remain limited in their ability to reason about safety, coherence, and trust across modalities. We introduce a reproducible simulation framework to evaluate generative agents in three aspects: (1) safety...