# Understanding Unreliability of Steering Vectors in Language Models: Geometric Predictors and the Limits of Linear Approximations

arXiv: https://arxiv.org/abs/2602.17881 · HTML: https://mlsanity.com/paper/2602.17881/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Activation steering methods, one of 13 directions in Mechanistic interpretability.
- First posted on arXiv 19 Feb 2026.

## Placed in

- Mechanistic interpretability / [Activation steering methods](https://mlsanity.com/mechinterp/activation-steering-methods/)

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

- [Exemplar Partitioning for Mechanistic Interpretability](https://mlsanity.com/paper/2605.14347/) (2026)
- [Activation Steering Transfer to Agents: One Gain Ratio Does Not Identify Potency and Efficacy](https://mlsanity.com/paper/2607.09156/) (2026)
- [Controlling Speaking Rate in Autoregressive TTS via Activation Steering](https://mlsanity.com/paper/2609.33810/) (2026)
- [Steering Follows Geometry, Not Labels: Emotion Directions in a Full-Duplex Speech Model](https://mlsanity.com/paper/2610.08887/) (2026)

## Abstract (arXiv)

Steering vectors are a lightweight method for controlling language model behavior by adding a learned bias to the activations at inference time. Although effective on average, steering effect sizes vary across samples and are unreliable for many target behaviors. In my thesis, I investigate why steering reliability differs across behaviors and how it is impacted by steering vector training data. First, I find that higher cosine similarity between training activation differences predicts more reliable steering. Second, I observe that behavior datasets where positive and negative activations are better separated along the steering direction are more reliably steerable. Finally, steering vectors trained on different prompt variations are directionally distinct, yet perform similarly well and exhibit correlated efficacy across datasets. My findings suggest that steering vectors are unreliable when the latent target behavior representation is not effectively approximated by the linear steering direction. Taken together, these insights offer a practical diagnostic for steering unreliability and motivate the development of more robust steering methods that explicitly account for non-linear latent behavior representations.