@inproceedings{youssef-etal-2025-fact,
title = "Has this Fact been Edited? Detecting Knowledge Edits in Language Models",
author = {Youssef, Paul and
Zhao, Zhixue and
Seifert, Christin and
Schl{\"o}tterer, J{\"o}rg},
editor = "Chiruzzo, Luis and
Ritter, Alan and
Wang, Lu",
booktitle = "Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)",
month = apr,
year = "2025",
address = "Albuquerque, New Mexico",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.naacl-long.492/",
pages = "9768--9784",
ISBN = "979-8-89176-189-6",
abstract = "Knowledge editing methods (KEs) can update language models' obsolete or inaccurate knowledge learned from pre-training. However, KEs can be used for malicious applications, e.g., inserting misinformation and toxic content. Knowing whether a generated output is based on edited knowledge or first-hand knowledge from pre-training can increase users' trust in generative models and provide more transparency. Driven by this, we propose a novel task: detecting knowledge edits in language models. Given an edited model and a fact retrieved by a prompt from an edited model, the objective is to classify the knowledge as either unedited (based on the pre-training), or edited (based on subsequent editing). We instantiate the task with four KEs, two large language models (LLMs), and two datasets. Additionally, we propose using hidden state representations and probability distributions as features for the detection model. Our results reveal that using these features as inputs to a simple AdaBoost classifier establishes a strong baseline. This baseline classifier requires a small amount of training data and maintains its performance even in cross-domain settings. Our work lays the groundwork for addressing potential malicious model editing, which is a critical challenge associated with the strong generative capabilities of LLMs."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="youssef-etal-2025-fact">
<titleInfo>
<title>Has this Fact been Edited? Detecting Knowledge Edits in Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Paul</namePart>
<namePart type="family">Youssef</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhixue</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Christin</namePart>
<namePart type="family">Seifert</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jörg</namePart>
<namePart type="family">Schlötterer</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-04</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Luis</namePart>
<namePart type="family">Chiruzzo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alan</namePart>
<namePart type="family">Ritter</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lu</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Albuquerque, New Mexico</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-189-6</identifier>
</relatedItem>
<abstract>Knowledge editing methods (KEs) can update language models’ obsolete or inaccurate knowledge learned from pre-training. However, KEs can be used for malicious applications, e.g., inserting misinformation and toxic content. Knowing whether a generated output is based on edited knowledge or first-hand knowledge from pre-training can increase users’ trust in generative models and provide more transparency. Driven by this, we propose a novel task: detecting knowledge edits in language models. Given an edited model and a fact retrieved by a prompt from an edited model, the objective is to classify the knowledge as either unedited (based on the pre-training), or edited (based on subsequent editing). We instantiate the task with four KEs, two large language models (LLMs), and two datasets. Additionally, we propose using hidden state representations and probability distributions as features for the detection model. Our results reveal that using these features as inputs to a simple AdaBoost classifier establishes a strong baseline. This baseline classifier requires a small amount of training data and maintains its performance even in cross-domain settings. Our work lays the groundwork for addressing potential malicious model editing, which is a critical challenge associated with the strong generative capabilities of LLMs.</abstract>
<identifier type="citekey">youssef-etal-2025-fact</identifier>
<location>
<url>https://aclanthology.org/2025.naacl-long.492/</url>
</location>
<part>
<date>2025-04</date>
<extent unit="page">
<start>9768</start>
<end>9784</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Has this Fact been Edited? Detecting Knowledge Edits in Language Models
%A Youssef, Paul
%A Zhao, Zhixue
%A Seifert, Christin
%A Schlötterer, Jörg
%Y Chiruzzo, Luis
%Y Ritter, Alan
%Y Wang, Lu
%S Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)
%D 2025
%8 April
%I Association for Computational Linguistics
%C Albuquerque, New Mexico
%@ 979-8-89176-189-6
%F youssef-etal-2025-fact
%X Knowledge editing methods (KEs) can update language models’ obsolete or inaccurate knowledge learned from pre-training. However, KEs can be used for malicious applications, e.g., inserting misinformation and toxic content. Knowing whether a generated output is based on edited knowledge or first-hand knowledge from pre-training can increase users’ trust in generative models and provide more transparency. Driven by this, we propose a novel task: detecting knowledge edits in language models. Given an edited model and a fact retrieved by a prompt from an edited model, the objective is to classify the knowledge as either unedited (based on the pre-training), or edited (based on subsequent editing). We instantiate the task with four KEs, two large language models (LLMs), and two datasets. Additionally, we propose using hidden state representations and probability distributions as features for the detection model. Our results reveal that using these features as inputs to a simple AdaBoost classifier establishes a strong baseline. This baseline classifier requires a small amount of training data and maintains its performance even in cross-domain settings. Our work lays the groundwork for addressing potential malicious model editing, which is a critical challenge associated with the strong generative capabilities of LLMs.
%U https://aclanthology.org/2025.naacl-long.492/
%P 9768-9784
Markdown (Informal)
[Has this Fact been Edited? Detecting Knowledge Edits in Language Models](https://aclanthology.org/2025.naacl-long.492/) (Youssef et al., NAACL 2025)
ACL
- Paul Youssef, Zhixue Zhao, Christin Seifert, and Jörg Schlötterer. 2025. Has this Fact been Edited? Detecting Knowledge Edits in Language Models. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers), pages 9768–9784, Albuquerque, New Mexico. Association for Computational Linguistics.