Refusal is a key safety behavior in aligned language models, yet the internal mechanisms driving refusals remain opaque. In this work, we conduct a mechanistic study of refusal in instruction-tuned LLMs using sparse autoencoders to identify latent features that causally mediate refusal behaviors. We apply our method to two open-source chat models and intervene on refusal-related features to assess their influence on generation, validating their behavioral impact across multiple harmful datasets. This enables a fine-grained inspection of how refusal manifests at the activation level and addresses key research questions such as investigating upstream-downstream latent relationship and understanding the mechanisms of adversarial jailbreaking techniques. We also establish the usefulness of refusal features in enhancing generalization for linear probes to out-of-distribution adversarial samples in classification tasks. We open source our code in https://github.com/wj210/refusal_sae.
@article{arxiv.2505.23556,
title = {Understanding Refusal in Language Models with Sparse Autoencoders},
author = {Wei Jie Yeo and Nirmalendu Prakash and Clement Neo and Roy Ka-Wei Lee and Erik Cambria and Ranjan Satapathy},
journal= {arXiv preprint arXiv:2505.23556},
year = {2025}
}