Persona-Model Collapse in Emergent Misalignment
cs.CL, cs.AI, cs.CR, cs.LG
Submitted: 2026-05-13
Updated: 2026-11-05
Code: https://github.com/bastoscostadavi/persona-moral-collapse-in-emergent-misalignment
Terminology
Sources
- Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs
- Emergent Misalignment via In-Context Learning: Narrow in-context examples can produce broadly misaligned LLMs
- Natural Emergent Misalignment from Reward Hacking in Production RL
- Thought Crime: Backdoors and Emergent Misalignment in Reasoning Models
- The Devil in the Details: Emergent Misalignment, Format and Coherence in Open-Weights LLMs
- Assessing Domain-Level Susceptibility to Emergent Misalignment from Narrow Finetuning
- Shared Parameter Subspaces and Cross-Task Linearity in Emergently Misaligned Behavior
- In-Training Defenses against Emergent Misalignment in Language Models
- Emergent misalignment as prompt sensitivity: A research note
- Moral Susceptibility and Robustness under Persona Role-Play in Large Language Models
- The political ideology of conversational AI: Converging evidence on ChatGPT's pro-environmental, left-libertarian orientation
- Differences in the Moral Foundations of Large Language Models
- Exploring and steering the moral compass of Large Language Models
- Are Large Language Models Moral Hypocrites? A Study Based on Moral Foundations
- Do Psychometric Tests Work for Large Language Models? Evaluation of Tests on Sexism, Racism, and Morality
- LoRA Fine-tuning Efficiently Undoes Safety Training in Llama 2-Chat 70B
- Assessing the Brittleness of Safety Alignment via Pruning and Low-Rank Modifications
- Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training
- Alignment faking in large language models
Related papers
- ALARM: Audio-Language Alignment for Reasoning Models
- ThinkSum: Probabilistic reasoning over sets using large language models
- Coherence boosting: When your pretrained language model is not paying enough attention
- From Facts to Insights: A Persona-Driven Dual Memory Framework and Dataset for Role-Playing Agents
- Improving General Role-Playing Agents via Psychology-Grounded Reasoning and Role-Aware Policy Optimization
- The Art of Asking: Multilingual Prompt Optimization for Synthetic Data