Metraipolis › archive › dc_f21223bbf17f
EmoRSS: Mitigating Emotion-Induced Over-Refusal in Large Language Models — arXiv:2610.04998
- Source
- arXiv
- Original
- https://arxiv.org/abs/2610.04998
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-06T05:24:56.299Z
- SHA-256 of the archived text
bdf8804814774fe7c77e97abe8238236f741ff32410bc9f160871030165524db- Stable URL of this page
https://metraipolis.com/documento/dc_f21223bbf17f
How to cite
EmoRSS: Mitigating Emotion-Induced Over-Refusal in Large Language Models — arXiv:2610.04998 — arXiv, https://arxiv.org/abs/2610.04998 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f21223bbf17f (Metraipolis archive dc_f21223bbf17f, sha256 bdf8804814774fe7c77e97abe8238236f741ff32410bc9f160871030165524db)
BibTeX
@misc{dcf21223bbf17f,
title = {EmoRSS: Mitigating Emotion-Induced Over-Refusal in Large Language Models — arXiv:2610.04998},
howpublished = {arXiv},
url = {https://arxiv.org/abs/2610.04998},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f21223bbf17f; sha256 bdf8804814774fe7c77e97abe8238236f741ff32410bc9f160871030165524db},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.04998. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Emotional expression can influence the safety decisions of large language models (LLMs), offering a potential avenue for improving safety alignment. Existing studies have mainly focused on how emotional expressions facilitate attacks under harmful requests, while overlooking their effects on benign requests. We find that emotional expression can also systematically increase refusal tendencies on benign requests, leading to unnecessary over-refusal. Based on this observation, we propose emotion-guided refusal subspace steering (EmoRSS), an activation-steering method that mitigates emotion-induced over-refusal while preserving refusal behaviour on harmful requests. Specifically, we first identify a refusal-sensitive layer using layer-wise linear probes and construct a refusal subspace from sparse autoencoder (SAE) features aligned with the probe direction. Next, we use paired regular and emotional requests with the same queries to estimate the mean activation shift in the features defining the refusal subspace. Finally, we decode this shift into an activation intervention vector and apply it in the reverse refusal direction during inference, without updating the backbone parameters. Experiments on two LLMs show that, when requests contain emotional expressions, our method achieves a more favourable trade-off between refusing harmful requests and answering benign ones than prior over-refusal mitigation baselines, while better preserving general task performance. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2610.04998 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-06T05:24:05.541Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).