# When Role-playing, Do Models Believe What They Say?

arXiv: https://arxiv.org/abs/2606.11502 · HTML: https://mlsanity.com/paper/2606.11502/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Builds on 22 of its 35 references that are in the map.
- First posted on arXiv 9 Jun 2026.

## Builds on (most cited first, 22 of 22)

- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [Inference-Time Intervention: Eliciting Truthful Answers from a Language Model](https://mlsanity.com/paper/2306.03341/) (2023)
- [DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence](https://mlsanity.com/paper/2606.19348/) (2026)
- [Discovering Latent Knowledge in Language Models Without Supervision](https://mlsanity.com/paper/2212.03827/) (2022)
- [The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasets](https://mlsanity.com/paper/2310.06824/) (2023)
- [The Reversal Curse: LLMs trained on "A is B" fail to learn "B is A"](https://mlsanity.com/paper/2309.12288/) (2023)
- [Role-Play with Large Language Models](https://mlsanity.com/paper/2305.16367/) (2023)
- [AI Deception: A Survey of Examples, Risks, and Potential Solutions](https://mlsanity.com/paper/2308.14752/) (2023)
- [Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs](https://mlsanity.com/paper/2502.17424/) (2025)
- [Mechanistically analyzing the effects of fine-tuning on procedurally defined tasks](https://mlsanity.com/paper/2311.12786/) (2023)
- [Model Organisms for Emergent Misalignment](https://mlsanity.com/paper/2506.11613/) (2025)
- [Weird Generalization and Inductive Backdoors: New Ways to Corrupt LLMs](https://mlsanity.com/paper/2512.09742/) (2025)
- [Inoculation Prompting: Eliciting traits from LLMs during training can suppress them at test-time](https://mlsanity.com/paper/2510.04340/) (2025)
- [No Answer Needed: Predicting LLM Answer Accuracy from Question-Only Linear Probes](https://mlsanity.com/paper/2509.10625/) (2025)
- [Alignment Pretraining: AI Discourse Causes Self-Fulfilling (Mis)alignment](https://mlsanity.com/paper/2601.10160/) (2026)
- [Language Models Represent Beliefs of Self and Others](https://mlsanity.com/paper/2402.18496/) (2024)
- [Believe It or Not: How Deeply do LLMs Believe Implanted Facts?](https://mlsanity.com/paper/2510.17941/) (2025)
- [Open Character Training: Shaping the Persona of AI Assistants through Constitutional AI](https://mlsanity.com/paper/2511.01689/) (2025)
- [Difficulties with Evaluating a Deception Detector for AIs](https://mlsanity.com/paper/2511.22662/) (2025)
- [Conditioning Predictive Models: Risks and Strategies](https://mlsanity.com/paper/2302.00805/) (2023)
- [Truth-value judgment in language models: 'truth directions' are context sensitive](https://mlsanity.com/paper/2404.18865/) (2024)
- [Tracing Persona Vectors Through LLM Pretraining](https://mlsanity.com/paper/2605.13329/) (2026)

## Built on by (most cited first, 4 of 4)

- [Character Training for Risk-Averse Agents](https://mlsanity.com/paper/2609.38093/) (2026)
- [Stress-Testing LLM Lie Detectors: Role-Play Failures and Spurious Correlations](https://mlsanity.com/paper/2609.39807/) (2026)
- [Pre-training interventions, ex post facto: Grafting model beliefs across checkpoints](https://mlsanity.com/paper/2610.00767/) (2026)
- [Predicting Alignment Generalization with Value Representations](https://mlsanity.com/paper/2610.12410/) (2026)

## Abstract (arXiv)

Language models can state that "the Earth orbits the Sun" and, when role-playing Aristotle, assert the opposite. Recent work argues that persona adoption is fundamental to how language models behave, with models selecting the most appropriate persona for a given context. Does such role-playing merely change the model's outputs, or does it also affect what the model internally represents as truthful? We study this question using the role-play of characters whose beliefs differ from the modern consensus, and induce personas with a number of different methods: prompting, in-context learning (ICL), supervised fine-tuning (SFT), and Open Character Training (OCT), and Emergent Misalignment (EM). We measure belief internalization across these approaches with truth probes and with behavioral tests, finding a broad spectrum of belief internalization. Prompting, ICL, and SFT change what the model says with little representational change. EM creates a large, broad shift in the model's truth representation, and OCT a smaller shift that is clearest on the larger model. Understanding when training changes a model's worldview rather than merely its behavior may become increasingly important as AI systems are entrusted with greater autonomy and influence.