# When Models Refuse: Political Steerability and Feature Richness as Measures of Ideological Depth

arXiv: https://arxiv.org/abs/2508.21448 · HTML: https://mlsanity.com/paper/2508.21448/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Safety alignment, refusal and guardrails, one of 12 directions in LLM safety and security.
- Builds on 13 of its 28 references that are in the map.
- First posted on arXiv 29 Aug 2025.

## Placed in

- LLM safety and security / [Safety alignment, refusal and guardrails](https://mlsanity.com/safety/safety-alignment-refusal-and-guardrails/)
- Mechanistic interpretability / [Probing deception, bias and beliefs](https://mlsanity.com/mechinterp/probing-deception-bias-and-beliefs/)

## Builds on (most cited first, 13 of 13)

- [Understanding intermediate layers using linear classifier probes](https://mlsanity.com/paper/1610.01644/) (2016)
- [Steering Llama 2 via Contrastive Activation Addition](https://mlsanity.com/paper/2312.06681/) (2023)
- [Semantics derived automatically from language corpora contain human-like biases](https://mlsanity.com/paper/1608.07187/) (2016)
- [Discovering Language Model Behaviors with Model-Written Evaluations](https://mlsanity.com/paper/2212.09251/) (2022)
- [Whose Opinions Do Language Models Reflect?](https://mlsanity.com/paper/2303.17548/) (2023)
- [Large Language Models Are Not Robust Multiple Choice Selectors](https://mlsanity.com/paper/2309.03882/) (2023)
- [Gemma Scope: Open Sparse Autoencoders Everywhere All At Once on Gemma 2](https://mlsanity.com/paper/2408.05147/) (2024)
- [Persona Vectors: Monitoring and Controlling Character Traits in Language Models](https://mlsanity.com/paper/2507.21509/) (2025)
- [Political Compass or Spinning Arrow? Towards More Meaningful Evaluations for Values and Opinions in Large Language Models](https://mlsanity.com/paper/2402.16786/) (2024)
- [Llama Scope: Extracting Millions of Features from Llama-3.1-8B with Sparse Autoencoders](https://mlsanity.com/paper/2410.20526/) (2024)
- [SAEs Are Good for Steering -- If You Select the Right Features](https://mlsanity.com/paper/2505.20063/) (2025)
- [Beyond Prompt Engineering: Robust Behavior Control in LLMs via Steering Target Atoms](https://mlsanity.com/paper/2505.20322/) (2025)
- [The Political Ideology of Large Language Models: Measurement, Inconsistency, and Persuasive Influence](https://mlsanity.com/paper/2505.04171/) (2025)

## Abstract (arXiv)

Large language models (LLMs) sometimes refuse to follow benign instructions, such as declining to argue a political position or adopt a stated persona, and such refusals are commonly read as safety guardrails at work. We ask whether they can instead signal a **capability deficit**: a shortage of the internal representations a model needs to reason from the instructed perspective. To investigate, we introduce **ideological depth**, a property with two components: (i) a model's ability to follow political instructions without *failure* (steerability), and (ii) the **feature richness** of its internal political representations, measured with sparse autoencoders (SAEs). Using two widely used openweight LLMs as candidates, we compare interventions based on prompts and activation-steering, and probe political features with publicly available SAEs. We find large, systematic differences: a model that is more steerable in both ideological directions activates **~7.3x** more distinct political features, while the other model instead responds with increased refusals. Causally ablating a small, targeted set of political features from the former model reproduces the same feature-poor behavior and drives up refusals. Together, these results indicate that refusals on benign prompts can arise from **capability deficits** rather than fixed safety rules, and that ideological depth is a measurable property of LLMs that helps predict when a model will refuse.