@inproceedings{jeung-etal-2026-dusk,
title = "{DUSK}: Do Not Unlearn Shared Knowledge",
author = "Jeung, Wonje and
Yoon, Sangyeon and
Hong, Hyesoo and
Kim, Soeun and
Han, Seungju and
Yu, Youngjae and
No, Albert",
editor = "Liakata, Maria and
Moreira, Viviane P. and
Zhang, Jiajun and
Jurgens, David",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
month = jul,
year = "2026",
address = "San Diego, California, United States",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-acl.2085/",
pages = "42013--42031",
ISBN = "979-8-89176-395-1",
abstract = "Machine unlearning aims to remove ``forget'' data while preserving knowledge from the ``retain'' data, yet a fundamental question arises when the two share content. By definition, an unlearned model should be indistinguishable from a model retrained solely on the retain set, which implies that shared knowledge must remain while only forget-specific content is removed. To evaluate this requirement, we introduce DUSK, the first benchmark for unlearning under realistic knowledge overlap. DUSK constructs documents containing both shared and unique knowledge and defines seven metrics to test whether methods erase forget-specific expressions without discarding shared facts. Evaluating nine recent approaches, we find that although surface text is often removed, current methods struggle to distinguish shared from unique knowledge, either erasing information that should be retained or failing to fully forget target content. DUSK provides a controlled, reproducible testbed for diagnosing these failures and guiding precise unlearning algorithms."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="jeung-etal-2026-dusk">
<titleInfo>
<title>DUSK: Do Not Unlearn Shared Knowledge</title>
</titleInfo>
<name type="personal">
<namePart type="given">Wonje</namePart>
<namePart type="family">Jeung</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sangyeon</namePart>
<namePart type="family">Yoon</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hyesoo</namePart>
<namePart type="family">Hong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Soeun</namePart>
<namePart type="family">Kim</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Seungju</namePart>
<namePart type="family">Han</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Youngjae</namePart>
<namePart type="family">Yu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Albert</namePart>
<namePart type="family">No</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: ACL 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Maria</namePart>
<namePart type="family">Liakata</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Viviane</namePart>
<namePart type="given">P</namePart>
<namePart type="family">Moreira</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiajun</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">David</namePart>
<namePart type="family">Jurgens</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">San Diego, California, United States</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-395-1</identifier>
</relatedItem>
<abstract>Machine unlearning aims to remove “forget” data while preserving knowledge from the “retain” data, yet a fundamental question arises when the two share content. By definition, an unlearned model should be indistinguishable from a model retrained solely on the retain set, which implies that shared knowledge must remain while only forget-specific content is removed. To evaluate this requirement, we introduce DUSK, the first benchmark for unlearning under realistic knowledge overlap. DUSK constructs documents containing both shared and unique knowledge and defines seven metrics to test whether methods erase forget-specific expressions without discarding shared facts. Evaluating nine recent approaches, we find that although surface text is often removed, current methods struggle to distinguish shared from unique knowledge, either erasing information that should be retained or failing to fully forget target content. DUSK provides a controlled, reproducible testbed for diagnosing these failures and guiding precise unlearning algorithms.</abstract>
<identifier type="citekey">jeung-etal-2026-dusk</identifier>
<location>
<url>https://aclanthology.org/2026.findings-acl.2085/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>42013</start>
<end>42031</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T DUSK: Do Not Unlearn Shared Knowledge
%A Jeung, Wonje
%A Yoon, Sangyeon
%A Hong, Hyesoo
%A Kim, Soeun
%A Han, Seungju
%A Yu, Youngjae
%A No, Albert
%Y Liakata, Maria
%Y Moreira, Viviane P.
%Y Zhang, Jiajun
%Y Jurgens, David
%S Findings of the Association for Computational Linguistics: ACL 2026
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, United States
%@ 979-8-89176-395-1
%F jeung-etal-2026-dusk
%X Machine unlearning aims to remove “forget” data while preserving knowledge from the “retain” data, yet a fundamental question arises when the two share content. By definition, an unlearned model should be indistinguishable from a model retrained solely on the retain set, which implies that shared knowledge must remain while only forget-specific content is removed. To evaluate this requirement, we introduce DUSK, the first benchmark for unlearning under realistic knowledge overlap. DUSK constructs documents containing both shared and unique knowledge and defines seven metrics to test whether methods erase forget-specific expressions without discarding shared facts. Evaluating nine recent approaches, we find that although surface text is often removed, current methods struggle to distinguish shared from unique knowledge, either erasing information that should be retained or failing to fully forget target content. DUSK provides a controlled, reproducible testbed for diagnosing these failures and guiding precise unlearning algorithms.
%U https://aclanthology.org/2026.findings-acl.2085/
%P 42013-42031
Markdown (Informal)
[DUSK: Do Not Unlearn Shared Knowledge](https://aclanthology.org/2026.findings-acl.2085/) (Jeung et al., Findings 2026)
ACL
- Wonje Jeung, Sangyeon Yoon, Hyesoo Hong, Soeun Kim, Seungju Han, Youngjae Yu, and Albert No. 2026. DUSK: Do Not Unlearn Shared Knowledge. In Findings of the Association for Computational Linguistics: ACL 2026, pages 42013–42031, San Diego, California, United States. Association for Computational Linguistics.