@inproceedings{chen-etal-2026-halumem,
title = "{H}alu{M}em: Evaluating Hallucinations in Memory Systems of Agents",
author = "Chen, Ding and
Niu, Simin and
Kehang, Li and
Liu, Peng and
Zheng, Xiangping and
Tang, Bo and
Li, Xinchi and
Xiong, Feiyu and
li, Zhiyu",
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.220/",
pages = "5073--5108",
ISBN = "979-8-89176-459-0",
abstract = "Memory systems are essential for enabling long-term learning and sustained interaction in AI systems such as LLMs and agents. However, memory storage and retrieval often suffer from hallucinations, including fabrication, errors, conflicts, and omissions. Existing evaluations mainly rely on end-to-end question answering, making it difficult to identify which memory operation causes hallucinations. To address this, we propose the Hallucination in Memory Benchmark (HaluMem), the first operation-level benchmark for evaluating hallucinations in memory systems. HaluMem defines three tasks (memory extraction, memory updating, and memory question answering) to comprehensively reveal hallucination behaviors across different operational stages of interaction. To support evaluation, we construct user-centric, multi-turn human-AI interaction datasets, HaluMem-Medium and HaluMem-Long, containing about 15k memory points and 3.5k questions. The datasets feature long dialogues, with average lengths of 1.5k and 2.6k turns and context sizes exceeding 1M tokens, enabling evaluation across varying context scales and task complexities. Experiments on HaluMem show that current memory systems tend to introduce and accumulate hallucinations during extraction and updating, which then propagate to question answering. These findings highlight the need for interpretable and constrained memory operation mechanisms to improve memory reliability. All resources are available at [GitHub](https://github.com/MemTensor/HaluMem)."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="chen-etal-2026-halumem">
<titleInfo>
<title>HaluMem: Evaluating Hallucinations in Memory Systems of Agents</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ding</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Simin</namePart>
<namePart type="family">Niu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Li</namePart>
<namePart type="family">Kehang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Peng</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiangping</namePart>
<namePart type="family">Zheng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bo</namePart>
<namePart type="family">Tang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xinchi</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Feiyu</namePart>
<namePart type="family">Xiong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhiyu</namePart>
<namePart type="family">li</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>Memory systems are essential for enabling long-term learning and sustained interaction in AI systems such as LLMs and agents. However, memory storage and retrieval often suffer from hallucinations, including fabrication, errors, conflicts, and omissions. Existing evaluations mainly rely on end-to-end question answering, making it difficult to identify which memory operation causes hallucinations. To address this, we propose the Hallucination in Memory Benchmark (HaluMem), the first operation-level benchmark for evaluating hallucinations in memory systems. HaluMem defines three tasks (memory extraction, memory updating, and memory question answering) to comprehensively reveal hallucination behaviors across different operational stages of interaction. To support evaluation, we construct user-centric, multi-turn human-AI interaction datasets, HaluMem-Medium and HaluMem-Long, containing about 15k memory points and 3.5k questions. The datasets feature long dialogues, with average lengths of 1.5k and 2.6k turns and context sizes exceeding 1M tokens, enabling evaluation across varying context scales and task complexities. Experiments on HaluMem show that current memory systems tend to introduce and accumulate hallucinations during extraction and updating, which then propagate to question answering. These findings highlight the need for interpretable and constrained memory operation mechanisms to improve memory reliability. All resources are available at [GitHub](https://github.com/MemTensor/HaluMem).</abstract>
<identifier type="citekey">chen-etal-2026-halumem</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.220/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>5073</start>
<end>5108</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T HaluMem: Evaluating Hallucinations in Memory Systems of Agents
%A Chen, Ding
%A Niu, Simin
%A Kehang, Li
%A Liu, Peng
%A Zheng, Xiangping
%A Tang, Bo
%A Li, Xinchi
%A Xiong, Feiyu
%A li, Zhiyu
%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-halumem
%X Memory systems are essential for enabling long-term learning and sustained interaction in AI systems such as LLMs and agents. However, memory storage and retrieval often suffer from hallucinations, including fabrication, errors, conflicts, and omissions. Existing evaluations mainly rely on end-to-end question answering, making it difficult to identify which memory operation causes hallucinations. To address this, we propose the Hallucination in Memory Benchmark (HaluMem), the first operation-level benchmark for evaluating hallucinations in memory systems. HaluMem defines three tasks (memory extraction, memory updating, and memory question answering) to comprehensively reveal hallucination behaviors across different operational stages of interaction. To support evaluation, we construct user-centric, multi-turn human-AI interaction datasets, HaluMem-Medium and HaluMem-Long, containing about 15k memory points and 3.5k questions. The datasets feature long dialogues, with average lengths of 1.5k and 2.6k turns and context sizes exceeding 1M tokens, enabling evaluation across varying context scales and task complexities. Experiments on HaluMem show that current memory systems tend to introduce and accumulate hallucinations during extraction and updating, which then propagate to question answering. These findings highlight the need for interpretable and constrained memory operation mechanisms to improve memory reliability. All resources are available at [GitHub](https://github.com/MemTensor/HaluMem).
%U https://aclanthology.org/2026.emnlp-main.220/
%P 5073-5108
Markdown (Informal)
[HaluMem: Evaluating Hallucinations in Memory Systems of Agents](https://aclanthology.org/2026.emnlp-main.220/) (Chen et al., EMNLP 2026)
ACL
- Ding Chen, Simin Niu, Li Kehang, Peng Liu, Xiangping Zheng, Bo Tang, Xinchi Li, Feiyu Xiong, and Zhiyu li. 2026. HaluMem: Evaluating Hallucinations in Memory Systems of Agents. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 5073–5108, Budapest, Hungary. Association for Computational Linguistics.