# Too Good to be Bad: On the Failure of LLMs to Role-Play Villains

arXiv: https://arxiv.org/abs/2511.04962 · HTML: https://mlsanity.com/paper/2511.04962/

## Key facts

- Cited by 11 papers in the map (data through 14 Oct 2026).
- Builds on 24 of its 34 references that are in the map.
- First posted on arXiv 7 Nov 2025.

## Builds on (most cited first, 24 of 24)

- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [GPT-4o System Card](https://mlsanity.com/paper/2410.21276/) (2024)
- [Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities](https://mlsanity.com/paper/2507.06261/) (2025)
- [Generative Agents: Interactive Simulacra of Human Behavior](https://mlsanity.com/paper/2304.03442/) (2023)
- [Fine-Tuning Language Models from Human Preferences](https://mlsanity.com/paper/1909.08593/) (2019)
- [Safe RLHF: Safe Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2310.12773/) (2023)
- [GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models](https://mlsanity.com/paper/2508.06471/) (2025)
- [GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher](https://mlsanity.com/paper/2308.06463/) (2023)
- [Persona Vectors: Monitoring and Controlling Character Traits in Language Models](https://mlsanity.com/paper/2507.21509/) (2025)
- [Pretraining Language Models with Human Preferences](https://mlsanity.com/paper/2302.08582/) (2023)
- [From Persona to Personalization: A Survey on Role-Playing Language Agents](https://mlsanity.com/paper/2404.18231/) (2024)
- [InCharacter: Evaluating Personality Fidelity in Role-Playing Agents through Psychological Interviews](https://mlsanity.com/paper/2310.17976/) (2023)
- [A Survey on Recent Advances in LLM-Based Multi-turn Dialogue Systems](https://mlsanity.com/paper/2402.18013/) (2024)
- [MiniMax-01: Scaling Foundation Models with Lightning Attention](https://mlsanity.com/paper/2501.08313/) (2025)
- [CharacterEval: A Chinese Benchmark for Role-Playing Conversational Agent Evaluation](https://mlsanity.com/paper/2401.01275/) (2024)
- [Safe LoRA: the Silver Lining of Reducing Safety Risks when Fine-tuning Large Language Models](https://mlsanity.com/paper/2405.16833/) (2024)
- [Know Your Limits: A Survey of Abstention in Large Language Models](https://mlsanity.com/paper/2407.18418/) (2024)
- [Refuse Whenever You Feel Unsafe: Improving Safety in LLMs via Decoupled Refusal Training](https://mlsanity.com/paper/2407.09121/) (2024)
- [SocialBench: Sociality Evaluation of Role-Playing Conversational Agents](https://mlsanity.com/paper/2403.13679/) (2024)
- [Hunyuan-TurboS: Advancing Large Language Models through Mamba-Transformer Synergy and Adaptive Chain-of-Thought](https://mlsanity.com/paper/2505.15431/) (2025)
- [Sentient Agent as a Judge: Evaluating Higher-Order Social Cognition in Large Language Models](https://mlsanity.com/paper/2505.02847/) (2025)
- [RAIDEN-R1: Improving Role-awareness of LLMs via GRPO with Verifiable Reward](https://mlsanity.com/paper/2505.10218/) (2025)
- [RPGBENCH: Evaluating Large Language Models as Role-Playing Game Engines](https://mlsanity.com/paper/2502.00595/) (2025)
- [BookWorld: From Novels to Interactive Agent Societies for Creative Story Generation](https://mlsanity.com/paper/2504.14538/) (2025)

## Built on by (most cited first, 11 of 11)

- [Self-Improving Pretraining: using post-trained models to pretrain better models](https://mlsanity.com/paper/2601.21343/) (2026)
- [Think Fast and Slow: Step-Level Cognitive Depth Adaptation for LLM Agents](https://mlsanity.com/paper/2602.12662/) (2026)
- [RoleCDE:Benchmarking and Mitigating Role-Alignment Trade-offs in Role-Playing Agents](https://mlsanity.com/paper/2606.01552/) (2026)
- [Identifying and Mitigating Bottlenecks in Role-Playing Agents: A Systematic Study of Disentangling Character Profile Axes](https://mlsanity.com/paper/2601.04716/) (2026)
- [Can LLMs Threaten Human Survival? Benchmarking Potential Existential Threats from LLMs via Prefix Completion](https://mlsanity.com/paper/2511.19171/) (2025)
- [Data Attribution of Emergent Misalignment with Persona Features](https://mlsanity.com/paper/2608.11025/) (2026)
- [PACIFIC: Can LLMs Discern the Psychometric Traits Influencing Your Preferences? Personality-Driven Preference Alignment in LLMs](https://mlsanity.com/paper/2602.07181/) (2026)
- [Absurd World: A Simple Yet Powerful Method to Absurdify the Real-world for Probing LLM Reasoning Capabilities](https://mlsanity.com/paper/2605.09678/) (2026)
- [The Storyteller in the Model: Narrative Pattern Inheritance, Escalation Dynamics, and Alignment Governance in LLMs](https://mlsanity.com/paper/2607.20449/) (2026)
- [Position: AI Is Not Ready for Strategic Conflicts](https://mlsanity.com/paper/2609.16189/) (2026)
- [ANIMASK: What the Model Contributes to Role Play in Simulated Story Worlds](https://mlsanity.com/paper/2609.16667/) (2026)

## Abstract (arXiv)

Large Language Models (LLMs) are increasingly tasked with creative generation, including the simulation of fictional characters. However, their ability to portray non-prosocial, antagonistic personas remains largely unexamined. We hypothesize that the safety alignment of modern LLMs creates a fundamental conflict with the task of authentically role-playing morally ambiguous or villainous characters. To investigate this, we introduce the Moral RolePlay benchmark, a new dataset featuring a four-level moral alignment scale and a balanced test set for rigorous evaluation. We task state-of-the-art LLMs with role-playing characters from moral paragons to pure villains. Our large-scale evaluation reveals a consistent, monotonic decline in role-playing fidelity as character morality decreases. We find that models struggle most with traits directly antithetical to safety principles, such as ``Deceitful'' and ``Manipulative'', often substituting nuanced malevolence with superficial aggression. Furthermore, we demonstrate that general chatbot proficiency is a poor predictor of villain role-playing ability, with highly safety-aligned models performing particularly poorly. Our work provides the first systematic evidence of this critical limitation, highlighting a key tension between model safety and creative fidelity. Our benchmark and findings pave the way for developing more nuanced, context-aware alignment methods.