Procedural content generation has enabled vast virtual worlds through levels, maps, and quests, but large-scale character generation remains underexplored. We identify two alignment-induced biases in existing methods: a positive moral bias, where characters uniformly adopt agreeable stances (e.g. always saying lying is bad), and a helpful assistant bias, where characters invariably answer questions directly (e.g. never refusing or deflecting). While such tendencies suit instruction-following systems, they suppress dramatic tension and yield predictable characters, stemming from maximum likelihood training and assistant fine-tuning. To address this, we introduce PersonaWeaver, a framework that disentangles world-building (roles, demographics) from behavioral-building (moral stances, interactional styles), yielding characters with more diverse reactions and moral stances, as well as second-order diversity in stylistic markers like length, tone, and punctuation. Code: https://github.com/mqraitem/Persona-Weaver
@article{arxiv.2601.03396,
title = {Breaking the Assistant Mold: Modeling Behavioral Variation in LLM Based Procedural Character Generation},
author = {Maan Qraitem and Kate Saenko and Bryan A. Plummer},
journal= {arXiv preprint arXiv:2601.03396},
year = {2026}
}