@inproceedings{mou-etal-2026-thinking,
title = "Thinking Twice Makes Large Language Models Safer and More Helpful",
author = "Mou, Yutao and
Luo, Yuxiao and
Zhang, Shikun and
Ye, Wei",
editor = "Liakata, Maria and
Moreira, Viviane P. and
Zhang, Jiajun and
Jurgens, David",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
month = jul,
year = "2026",
address = "San Diego, California, United States",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-acl.1812/",
pages = "36365--36389",
ISBN = "979-8-89176-395-1",
abstract = "Current safety alignment techniques for large language models (LLMs) struggle to balance harmlessness and helpfulness: improving safety often comes at the cost of degraded utility. Our preliminary study shows that guiding unaligned base models with safety-aware reasoning that includes explicit self-reflection can effectively defend jailbreak attacks while preserving response quality. This observation motivates internalizing and strengthening self-reflective reasoning capabilities within LLMs to achieve a better safety{--}utility trade-off. We propose Safety-aware Reflective Reasoning Optimization (SaRO), a two-stage framework: (1) Reasoning-style Warmup (RW) to internalize self-reflective reasoning, and (2) Self-reflective Reasoning Process Optimization (SRPO) to encourage reflection and correction. Experiments show that SaRO outperforms existing reasoning-based alignment methods, achieving a better balance of safety and helpfulness."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="mou-etal-2026-thinking">
<titleInfo>
<title>Thinking Twice Makes Large Language Models Safer and More Helpful</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yutao</namePart>
<namePart type="family">Mou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yuxiao</namePart>
<namePart type="family">Luo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shikun</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wei</namePart>
<namePart type="family">Ye</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: ACL 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Maria</namePart>
<namePart type="family">Liakata</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Viviane</namePart>
<namePart type="given">P</namePart>
<namePart type="family">Moreira</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiajun</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">David</namePart>
<namePart type="family">Jurgens</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">San Diego, California, United States</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-395-1</identifier>
</relatedItem>
<abstract>Current safety alignment techniques for large language models (LLMs) struggle to balance harmlessness and helpfulness: improving safety often comes at the cost of degraded utility. Our preliminary study shows that guiding unaligned base models with safety-aware reasoning that includes explicit self-reflection can effectively defend jailbreak attacks while preserving response quality. This observation motivates internalizing and strengthening self-reflective reasoning capabilities within LLMs to achieve a better safety–utility trade-off. We propose Safety-aware Reflective Reasoning Optimization (SaRO), a two-stage framework: (1) Reasoning-style Warmup (RW) to internalize self-reflective reasoning, and (2) Self-reflective Reasoning Process Optimization (SRPO) to encourage reflection and correction. Experiments show that SaRO outperforms existing reasoning-based alignment methods, achieving a better balance of safety and helpfulness.</abstract>
<identifier type="citekey">mou-etal-2026-thinking</identifier>
<location>
<url>https://aclanthology.org/2026.findings-acl.1812/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>36365</start>
<end>36389</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Thinking Twice Makes Large Language Models Safer and More Helpful
%A Mou, Yutao
%A Luo, Yuxiao
%A Zhang, Shikun
%A Ye, Wei
%Y Liakata, Maria
%Y Moreira, Viviane P.
%Y Zhang, Jiajun
%Y Jurgens, David
%S Findings of the Association for Computational Linguistics: ACL 2026
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, United States
%@ 979-8-89176-395-1
%F mou-etal-2026-thinking
%X Current safety alignment techniques for large language models (LLMs) struggle to balance harmlessness and helpfulness: improving safety often comes at the cost of degraded utility. Our preliminary study shows that guiding unaligned base models with safety-aware reasoning that includes explicit self-reflection can effectively defend jailbreak attacks while preserving response quality. This observation motivates internalizing and strengthening self-reflective reasoning capabilities within LLMs to achieve a better safety–utility trade-off. We propose Safety-aware Reflective Reasoning Optimization (SaRO), a two-stage framework: (1) Reasoning-style Warmup (RW) to internalize self-reflective reasoning, and (2) Self-reflective Reasoning Process Optimization (SRPO) to encourage reflection and correction. Experiments show that SaRO outperforms existing reasoning-based alignment methods, achieving a better balance of safety and helpfulness.
%U https://aclanthology.org/2026.findings-acl.1812/
%P 36365-36389
Markdown (Informal)
[Thinking Twice Makes Large Language Models Safer and More Helpful](https://aclanthology.org/2026.findings-acl.1812/) (Mou et al., Findings 2026)
ACL