@inproceedings{chen-etal-2026-dont,
title = "Don{'}t Start from Scratch: Memory-Informed Detoxification for Safer {LLM} Responses",
author = "Chen, Pengze and
Wang, Jiachuan and
Sun, Weijie and
Ni, Wangze and
Chen, Lei",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.2452/",
pages = "52627--52644",
ISBN = "979-8-89176-462-0",
abstract = "Large language models demonstrate exceptional capabilities but raise critical safety concerns due to toxic outputs causing potential harm to users. Prompt optimization has emerged as a practical detoxification solution because it operates at the input level and can be applied to black-box or API-based models without parameter access. However, existing methods treat each task in a stateless manner without leveraging prior optimization experiences, which results in either ineffective detoxification or over-sanitized prompts that lose original intent in the vast search space. To address this, we propose the Memory-INformed Detoxification (MIND) framework that harnesses historical optimization experiences to enhance detoxification effectiveness through Memory Retrieval (MR) for retrieving relevant past cases and Memory Maintenance (MM) for maintaining a representative memory pool. We formulate MR and MM as optimization problems, prove their NP-hardness, and develop two approximation algorithms with constant approximation ratios for these problems. Experiments across multiple datasets and model configurations show that MIND outperforms existing baselines in both toxicity reduction and response utility."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="chen-etal-2026-dont">
<titleInfo>
<title>Don’t Start from Scratch: Memory-Informed Detoxification for Safer LLM Responses</title>
</titleInfo>
<name type="personal">
<namePart type="given">Pengze</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiachuan</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Weijie</namePart>
<namePart type="family">Sun</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wangze</namePart>
<namePart type="family">Ni</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lei</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Large language models demonstrate exceptional capabilities but raise critical safety concerns due to toxic outputs causing potential harm to users. Prompt optimization has emerged as a practical detoxification solution because it operates at the input level and can be applied to black-box or API-based models without parameter access. However, existing methods treat each task in a stateless manner without leveraging prior optimization experiences, which results in either ineffective detoxification or over-sanitized prompts that lose original intent in the vast search space. To address this, we propose the Memory-INformed Detoxification (MIND) framework that harnesses historical optimization experiences to enhance detoxification effectiveness through Memory Retrieval (MR) for retrieving relevant past cases and Memory Maintenance (MM) for maintaining a representative memory pool. We formulate MR and MM as optimization problems, prove their NP-hardness, and develop two approximation algorithms with constant approximation ratios for these problems. Experiments across multiple datasets and model configurations show that MIND outperforms existing baselines in both toxicity reduction and response utility.</abstract>
<identifier type="citekey">chen-etal-2026-dont</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.2452/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>52627</start>
<end>52644</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Don’t Start from Scratch: Memory-Informed Detoxification for Safer LLM Responses
%A Chen, Pengze
%A Wang, Jiachuan
%A Sun, Weijie
%A Ni, Wangze
%A Chen, Lei
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F chen-etal-2026-dont
%X Large language models demonstrate exceptional capabilities but raise critical safety concerns due to toxic outputs causing potential harm to users. Prompt optimization has emerged as a practical detoxification solution because it operates at the input level and can be applied to black-box or API-based models without parameter access. However, existing methods treat each task in a stateless manner without leveraging prior optimization experiences, which results in either ineffective detoxification or over-sanitized prompts that lose original intent in the vast search space. To address this, we propose the Memory-INformed Detoxification (MIND) framework that harnesses historical optimization experiences to enhance detoxification effectiveness through Memory Retrieval (MR) for retrieving relevant past cases and Memory Maintenance (MM) for maintaining a representative memory pool. We formulate MR and MM as optimization problems, prove their NP-hardness, and develop two approximation algorithms with constant approximation ratios for these problems. Experiments across multiple datasets and model configurations show that MIND outperforms existing baselines in both toxicity reduction and response utility.
%U https://aclanthology.org/2026.findings-emnlp.2452/
%P 52627-52644
Markdown (Informal)
[Don’t Start from Scratch: Memory-Informed Detoxification for Safer LLM Responses](https://aclanthology.org/2026.findings-emnlp.2452/) (Chen et al., Findings 2026)
ACL