# ChatGPT Reads Your Tone and Responds Accordingly -- Until It Does Not -- Emotional Framing Induces Bias in LLM Outputs

arXiv: https://arxiv.org/abs/2507.21083 · HTML: https://mlsanity.com/paper/2507.21083/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Builds on 3 of its 5 references that are in the map.
- First posted on arXiv 17 Jun 2025.

## Builds on (most cited first, 3 of 3)

- [Discovering Language Model Behaviors with Model-Written Evaluations](https://mlsanity.com/paper/2212.09251/) (2022)
- [Large Language Models Understand and Can be Enhanced by Emotional Stimuli](https://mlsanity.com/paper/2307.11760/) (2023)
- [Emotional Manipulation Through Prompt Engineering Amplifies Disinformation Generation in AI Large Language Models](https://mlsanity.com/paper/2403.03550/) (2024)

## Built on by (most cited first, 6 of 6)

- [Too Open for Opinion? Embracing Open-Endedness in Large Language Models for Social Simulation](https://mlsanity.com/paper/2510.13884/) (2025)
- [Bias Beneath the Tone: Empirical Characterisation of Tone Bias in LLM-Driven UX Systems](https://mlsanity.com/paper/2512.19950/) (2025)
- [Evaluating the Usage of African-American Vernacular English in Large Language Models](https://mlsanity.com/paper/2602.21485/) (2026)
- [LLMs Should Incorporate Explicit Mechanisms for Human Empathy](https://mlsanity.com/paper/2604.10557/) (2026)
- [Defeat Devices in AI Systems](https://mlsanity.com/paper/2606.28863/) (2026)
- [Reporting Under Pressure: Separating Factual and Tonal Sycophancy in LLM Statistical Analysis](https://mlsanity.com/paper/2609.27756/) (2026)

## Abstract (arXiv)

Large Language Models like GPT-4 adjust their responses not only based on the question asked, but also on how it is emotionally phrased. We systematically vary the emotional tone of 156 prompts - spanning controversial and everyday topics - and analyze how it affects model responses. Our findings show that GPT-4 is three times less likely to respond negatively to a negatively framed question than to a neutral one. This suggests a "rebound" bias where the model overcorrects, often shifting toward neutrality or positivity. On sensitive topics (e.g., justice or politics), this effect is even more pronounced: tone-based variation is suppressed, suggesting an alignment override. We introduce concepts like the "tone floor" - a lower bound in response negativity - and use tone-valence transition matrices to quantify behavior. Visualizations based on 1536-dimensional embeddings confirm semantic drift based on tone. Our work highlights an underexplored class of biases driven by emotional framing in prompts, with implications for AI alignment and trust. Code and data are available at: https://github.com/bardolfranck/llm-responses-viewer