@inproceedings{tan-etal-2026-fact,
title = "Fact-Checking the Jailbreak: A Claim-Level Framework for Structured Misinformation Assessment",
author = "Tan, Jia Yee and
Fu, Rong and
Liu, Yongtai and
Hu, Kaiyi and
Fong, Simon James",
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.23/",
pages = "429--448",
ISBN = "979-8-89176-462-0",
abstract = "Large language models (LLMs) remain vulnerable to jailbreak attacks that elicit structured, credible misinformation. Existing benchmarks primarily track binary refusals, often overlooking the severity and exposure risk of the misinformation payload itself. In this work, we introduce Misinfo-Jailbreak Bench, a comprehensive framework that systematically maps diverse attack strategies to a claim-level scoring pipeline for precise risk quantification. Building on this diagnostic foundation, we propose Selective LoRA Ethical Surgery, a parameter-efficient defense mechanism. Instead of broad parameter updates, our approach identifies safety-relevant subspaces via contrastive representations to confine LoRA edits and employs inference-time gating to preserve benign capabilities. Experimental results on state-of-the-art models (e.g., Qwen2.5, Llama-4) demonstrate that our method reduces high-severity risk by 60-70{\%} while limiting general capability degradation to less than 1{\%}, offering a precise and minimally invasive paradigm for post-hoc alignment."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="tan-etal-2026-fact">
<titleInfo>
<title>Fact-Checking the Jailbreak: A Claim-Level Framework for Structured Misinformation Assessment</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jia</namePart>
<namePart type="given">Yee</namePart>
<namePart type="family">Tan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rong</namePart>
<namePart type="family">Fu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yongtai</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kaiyi</namePart>
<namePart type="family">Hu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Simon</namePart>
<namePart type="given">James</namePart>
<namePart type="family">Fong</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 (LLMs) remain vulnerable to jailbreak attacks that elicit structured, credible misinformation. Existing benchmarks primarily track binary refusals, often overlooking the severity and exposure risk of the misinformation payload itself. In this work, we introduce Misinfo-Jailbreak Bench, a comprehensive framework that systematically maps diverse attack strategies to a claim-level scoring pipeline for precise risk quantification. Building on this diagnostic foundation, we propose Selective LoRA Ethical Surgery, a parameter-efficient defense mechanism. Instead of broad parameter updates, our approach identifies safety-relevant subspaces via contrastive representations to confine LoRA edits and employs inference-time gating to preserve benign capabilities. Experimental results on state-of-the-art models (e.g., Qwen2.5, Llama-4) demonstrate that our method reduces high-severity risk by 60-70% while limiting general capability degradation to less than 1%, offering a precise and minimally invasive paradigm for post-hoc alignment.</abstract>
<identifier type="citekey">tan-etal-2026-fact</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.23/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>429</start>
<end>448</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Fact-Checking the Jailbreak: A Claim-Level Framework for Structured Misinformation Assessment
%A Tan, Jia Yee
%A Fu, Rong
%A Liu, Yongtai
%A Hu, Kaiyi
%A Fong, Simon James
%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 tan-etal-2026-fact
%X Large language models (LLMs) remain vulnerable to jailbreak attacks that elicit structured, credible misinformation. Existing benchmarks primarily track binary refusals, often overlooking the severity and exposure risk of the misinformation payload itself. In this work, we introduce Misinfo-Jailbreak Bench, a comprehensive framework that systematically maps diverse attack strategies to a claim-level scoring pipeline for precise risk quantification. Building on this diagnostic foundation, we propose Selective LoRA Ethical Surgery, a parameter-efficient defense mechanism. Instead of broad parameter updates, our approach identifies safety-relevant subspaces via contrastive representations to confine LoRA edits and employs inference-time gating to preserve benign capabilities. Experimental results on state-of-the-art models (e.g., Qwen2.5, Llama-4) demonstrate that our method reduces high-severity risk by 60-70% while limiting general capability degradation to less than 1%, offering a precise and minimally invasive paradigm for post-hoc alignment.
%U https://aclanthology.org/2026.findings-emnlp.23/
%P 429-448
Markdown (Informal)
[Fact-Checking the Jailbreak: A Claim-Level Framework for Structured Misinformation Assessment](https://aclanthology.org/2026.findings-emnlp.23/) (Tan et al., Findings 2026)
ACL