@inproceedings{chen-etal-2026-visa,
title = "{VISA}: Value Injection via Shielded Adaptation for Personalized {LLM} Alignment",
author = "Chen, Jiawei and
Yang, Tianzhuo and
Zhang, Guoxi and
Ji, Jiaming and
Yang, Yaodong and
Dai, Juntao",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.303/",
pages = "6973--6999",
ISBN = "979-8-89176-459-0",
abstract = "Personalizing large language models often requires adapting them to new domains, tasks, or user preferences. However, such adaptation can unintentionally shift a model{'}s value profile, even when the training data is primarily knowledge-oriented. We refer to this failure mode as \textit{value drift}, a form of value alignment tax that emerges when learning new knowledge interferes with pre-existing alignment. To address this problem, we propose VISA, a modular framework for Value Injection via Shielded Adaptation. VISA keeps the base model as a stable content generator and trains a lightweight Value Rewriter to adjust the value expression of generated responses toward a target value vector. The target is constructed through a Value Detector and an Instruction Translator, while the rewriter is optimized with Group Relative Policy Optimization using rewards for value alignment and semantic consistency. We further construct VCR-45K, a value-conditioned rewriting dataset for evaluating controllable value adaptation. Experiments show that VISA substantially improves semantic preservation over prompting and fine-tuning baselines while maintaining competitive value alignment, enabling low-cost personalized alignment with reduced value drift."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="chen-etal-2026-visa">
<titleInfo>
<title>VISA: Value Injection via Shielded Adaptation for Personalized LLM Alignment</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jiawei</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tianzhuo</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Guoxi</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiaming</namePart>
<namePart type="family">Ji</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yaodong</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Juntao</namePart>
<namePart type="family">Dai</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>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</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">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Personalizing large language models often requires adapting them to new domains, tasks, or user preferences. However, such adaptation can unintentionally shift a model’s value profile, even when the training data is primarily knowledge-oriented. We refer to this failure mode as value drift, a form of value alignment tax that emerges when learning new knowledge interferes with pre-existing alignment. To address this problem, we propose VISA, a modular framework for Value Injection via Shielded Adaptation. VISA keeps the base model as a stable content generator and trains a lightweight Value Rewriter to adjust the value expression of generated responses toward a target value vector. The target is constructed through a Value Detector and an Instruction Translator, while the rewriter is optimized with Group Relative Policy Optimization using rewards for value alignment and semantic consistency. We further construct VCR-45K, a value-conditioned rewriting dataset for evaluating controllable value adaptation. Experiments show that VISA substantially improves semantic preservation over prompting and fine-tuning baselines while maintaining competitive value alignment, enabling low-cost personalized alignment with reduced value drift.</abstract>
<identifier type="citekey">chen-etal-2026-visa</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.303/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>6973</start>
<end>6999</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T VISA: Value Injection via Shielded Adaptation for Personalized LLM Alignment
%A Chen, Jiawei
%A Yang, Tianzhuo
%A Zhang, Guoxi
%A Ji, Jiaming
%A Yang, Yaodong
%A Dai, Juntao
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F chen-etal-2026-visa
%X Personalizing large language models often requires adapting them to new domains, tasks, or user preferences. However, such adaptation can unintentionally shift a model’s value profile, even when the training data is primarily knowledge-oriented. We refer to this failure mode as value drift, a form of value alignment tax that emerges when learning new knowledge interferes with pre-existing alignment. To address this problem, we propose VISA, a modular framework for Value Injection via Shielded Adaptation. VISA keeps the base model as a stable content generator and trains a lightweight Value Rewriter to adjust the value expression of generated responses toward a target value vector. The target is constructed through a Value Detector and an Instruction Translator, while the rewriter is optimized with Group Relative Policy Optimization using rewards for value alignment and semantic consistency. We further construct VCR-45K, a value-conditioned rewriting dataset for evaluating controllable value adaptation. Experiments show that VISA substantially improves semantic preservation over prompting and fine-tuning baselines while maintaining competitive value alignment, enabling low-cost personalized alignment with reduced value drift.
%U https://aclanthology.org/2026.emnlp-main.303/
%P 6973-6999
Markdown (Informal)
[VISA: Value Injection via Shielded Adaptation for Personalized LLM Alignment](https://aclanthology.org/2026.emnlp-main.303/) (Chen et al., EMNLP 2026)
ACL