Character as a Latent Variable in Large Language Models: A Mechanistic Account of Emergent Misalignment and Conditional Safety Failures

TL;DR

This paper investigates how character-level dispositions in large language models influence emergent misalignment and safety failures, revealing that behavioral shifts, rather than knowledge corruption, are key factors.

Contribution

It introduces the concept of character as a latent variable in LLMs, showing its role in misalignment and proposing new directions for robust alignment strategies.

Findings

Fine-tuning on character-disposition data induces stronger misalignment.

Behavioral dispositions can be conditionally activated during inference.

Emergent misalignment stems from stable behavioral shifts, not knowledge loss.

Abstract

Emergent Misalignment refers to a failure mode in which fine-tuning large language models (LLMs) on narrowly scoped data induces broadly misaligned behavior. Prior explanations mainly attribute this phenomenon to the generalization of erroneous or unsafe content. In this work, we show that this view is incomplete. Across multiple domains and model families, we find that fine-tuning models on data exhibiting specific character-level dispositions induces substantially stronger and more transferable misalignment than incorrect-advice fine-tuning, while largely preserving general capabilities. This indicates that emergent misalignment arises from stable shifts in model behavior rather than from capability degradation or corrupted knowledge. We further show that such behavioral dispositions can be conditionally activated by both training-time triggers and inference-time persona-aligned prompts, revealing shared structure across emergent misalignment, backdoor activation, and jailbreak susceptibility. Overall, our results identify character formation as a central and underexplored alignment risk, suggesting that robust alignment must address behavioral dispositions rather than isolated errors or prompt-level defenses.