<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="meade-etal-2025-investigating">
    <titleInfo>
        <title>Investigating Adversarial Trigger Transfer in Large Language Models</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Nicholas</namePart>
        <namePart type="family">Meade</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Arkil</namePart>
        <namePart type="family">Patel</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Siva</namePart>
        <namePart type="family">Reddy</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2025</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <genre authority="bibutilsgt">journal article</genre>
    <relatedItem type="host">
        <titleInfo>
            <title>Transactions of the Association for Computational Linguistics</title>
        </titleInfo>
        <originInfo>
            <issuance>continuing</issuance>
            <publisher>MIT Press</publisher>
            <place>
                <placeTerm type="text">Cambridge, MA</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">periodical</genre>
        <genre authority="bibutilsgt">academic journal</genre>
    </relatedItem>
    <abstract>Recent work has developed optimization procedures to find token sequences, called adversarial triggers, which can elicit unsafe responses from aligned language models. These triggers are believed to be highly transferable, i.e., a trigger optimized on one model can jailbreak other models. In this paper, we concretely show that such adversarial triggers are not consistently transferable. We extensively investigate trigger transfer among 13 open models and observe poor and inconsistent transfer. Our experiments further reveal a significant difference in robustness to adversarial triggers between models Aligned by Preference Optimization (APO) and models Aligned by Fine-Tuning (AFT). We find that APO models are extremely hard to jailbreak even when the trigger is optimized directly on the model. On the other hand, while AFT models may appear safe on the surface, exhibiting refusals to a range of unsafe instructions, we show that they are highly susceptible to adversarial triggers. Lastly, we observe that most triggers optimized on AFT models also generalize to new unsafe instructions from five diverse domains, further emphasizing their vulnerability. Overall, our work highlights the need for more comprehensive safety evaluations for aligned language models.1 Warning: This paper contains examples that may be offensive or upsetting.</abstract>
    <identifier type="citekey">meade-etal-2025-investigating</identifier>
    <identifier type="doi">10.1162/tacl.a.27</identifier>
    <location>
        <url>https://aclanthology.org/2025.tacl-1.45/</url>
    </location>
    <part>
        <date>2025</date>
        <detail type="volume"><number>13</number></detail>
        <extent unit="page">
            <start>953</start>
            <end>979</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Journal Article
%T Investigating Adversarial Trigger Transfer in Large Language Models
%A Meade, Nicholas
%A Patel, Arkil
%A Reddy, Siva
%J Transactions of the Association for Computational Linguistics
%D 2025
%V 13
%I MIT Press
%C Cambridge, MA
%F meade-etal-2025-investigating
%X Recent work has developed optimization procedures to find token sequences, called adversarial triggers, which can elicit unsafe responses from aligned language models. These triggers are believed to be highly transferable, i.e., a trigger optimized on one model can jailbreak other models. In this paper, we concretely show that such adversarial triggers are not consistently transferable. We extensively investigate trigger transfer among 13 open models and observe poor and inconsistent transfer. Our experiments further reveal a significant difference in robustness to adversarial triggers between models Aligned by Preference Optimization (APO) and models Aligned by Fine-Tuning (AFT). We find that APO models are extremely hard to jailbreak even when the trigger is optimized directly on the model. On the other hand, while AFT models may appear safe on the surface, exhibiting refusals to a range of unsafe instructions, we show that they are highly susceptible to adversarial triggers. Lastly, we observe that most triggers optimized on AFT models also generalize to new unsafe instructions from five diverse domains, further emphasizing their vulnerability. Overall, our work highlights the need for more comprehensive safety evaluations for aligned language models.1 Warning: This paper contains examples that may be offensive or upsetting.
%R 10.1162/tacl.a.27
%U https://aclanthology.org/2025.tacl-1.45/
%U https://doi.org/10.1162/tacl.a.27
%P 953-979

Markdown (Informal)

[Investigating Adversarial Trigger Transfer in Large Language Models](https://aclanthology.org/2025.tacl-1.45/) (Meade et al., TACL 2025)

• Investigating Adversarial Trigger Transfer in Large Language Models (Meade et al., TACL 2025)

ACL

• Nicholas Meade, Arkil Patel, and Siva Reddy. 2025. Investigating Adversarial Trigger Transfer in Large Language Models. Transactions of the Association for Computational Linguistics, 13:953–979.