@inproceedings{ivanov-etal-2026-toolhace,
title = "{T}ool{HACE}: A Dataset for Detecting Hallucinations in Tool-Based {LLM} Responses",
author = "Ivanov, Nikolay and
Pletenev, Sergey and
Rykov, Elisei and
Anatolii, Frolov and
Monogov, Bogdan and
Potekhin, Alexander and
Marina, Maria and
Moskovskiy, Daniil and
Moskvoretskii, Viktor and
Panchenko, Alexander and
Alimova, Ilseyar",
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.1919/",
pages = "40915--40934",
ISBN = "979-8-89176-462-0",
abstract = "Tool-augmented language models ground their responses in external tool outputs, improving reliability beyond parametric knowledge alone. However, even when the correct tool is called and returns accurate information, models may still contradict such outputs, omit critical details, or generate unsupported claims. Existing hallucination benchmarks for tool-augmented dialogue primarily focus on the tool-calling stage, evaluating whether the correct tool and arguments are selected, while overlooking errors that arise during post-tool response generation. To address this gap, we introduce , a span-level benchmark for hallucination detection in tool-augmented LLM responses, covering five fine-grained hallucination categories. Through extensive experiments with modern hallucination detectors, we show that post-tool-call hallucinations are widespread and remain challenging to detect. Our results further demonstrate that effective detection requires training on specialized tool-augmented corpora, while transfer from tool-agnostic hallucination datasets performs poorly in this setting. These findings highlight the need for dedicated benchmarks and models for reliable hallucination detection in tool-augmented language systems."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="ivanov-etal-2026-toolhace">
<titleInfo>
<title>ToolHACE: A Dataset for Detecting Hallucinations in Tool-Based LLM Responses</title>
</titleInfo>
<name type="personal">
<namePart type="given">Nikolay</namePart>
<namePart type="family">Ivanov</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sergey</namePart>
<namePart type="family">Pletenev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Elisei</namePart>
<namePart type="family">Rykov</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Frolov</namePart>
<namePart type="family">Anatolii</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bogdan</namePart>
<namePart type="family">Monogov</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alexander</namePart>
<namePart type="family">Potekhin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Maria</namePart>
<namePart type="family">Marina</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Daniil</namePart>
<namePart type="family">Moskovskiy</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Viktor</namePart>
<namePart type="family">Moskvoretskii</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alexander</namePart>
<namePart type="family">Panchenko</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ilseyar</namePart>
<namePart type="family">Alimova</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>Tool-augmented language models ground their responses in external tool outputs, improving reliability beyond parametric knowledge alone. However, even when the correct tool is called and returns accurate information, models may still contradict such outputs, omit critical details, or generate unsupported claims. Existing hallucination benchmarks for tool-augmented dialogue primarily focus on the tool-calling stage, evaluating whether the correct tool and arguments are selected, while overlooking errors that arise during post-tool response generation. To address this gap, we introduce , a span-level benchmark for hallucination detection in tool-augmented LLM responses, covering five fine-grained hallucination categories. Through extensive experiments with modern hallucination detectors, we show that post-tool-call hallucinations are widespread and remain challenging to detect. Our results further demonstrate that effective detection requires training on specialized tool-augmented corpora, while transfer from tool-agnostic hallucination datasets performs poorly in this setting. These findings highlight the need for dedicated benchmarks and models for reliable hallucination detection in tool-augmented language systems.</abstract>
<identifier type="citekey">ivanov-etal-2026-toolhace</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1919/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>40915</start>
<end>40934</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T ToolHACE: A Dataset for Detecting Hallucinations in Tool-Based LLM Responses
%A Ivanov, Nikolay
%A Pletenev, Sergey
%A Rykov, Elisei
%A Anatolii, Frolov
%A Monogov, Bogdan
%A Potekhin, Alexander
%A Marina, Maria
%A Moskovskiy, Daniil
%A Moskvoretskii, Viktor
%A Panchenko, Alexander
%A Alimova, Ilseyar
%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 ivanov-etal-2026-toolhace
%X Tool-augmented language models ground their responses in external tool outputs, improving reliability beyond parametric knowledge alone. However, even when the correct tool is called and returns accurate information, models may still contradict such outputs, omit critical details, or generate unsupported claims. Existing hallucination benchmarks for tool-augmented dialogue primarily focus on the tool-calling stage, evaluating whether the correct tool and arguments are selected, while overlooking errors that arise during post-tool response generation. To address this gap, we introduce , a span-level benchmark for hallucination detection in tool-augmented LLM responses, covering five fine-grained hallucination categories. Through extensive experiments with modern hallucination detectors, we show that post-tool-call hallucinations are widespread and remain challenging to detect. Our results further demonstrate that effective detection requires training on specialized tool-augmented corpora, while transfer from tool-agnostic hallucination datasets performs poorly in this setting. These findings highlight the need for dedicated benchmarks and models for reliable hallucination detection in tool-augmented language systems.
%U https://aclanthology.org/2026.findings-emnlp.1919/
%P 40915-40934
Markdown (Informal)
[ToolHACE: A Dataset for Detecting Hallucinations in Tool-Based LLM Responses](https://aclanthology.org/2026.findings-emnlp.1919/) (Ivanov et al., Findings 2026)
ACL
- Nikolay Ivanov, Sergey Pletenev, Elisei Rykov, Frolov Anatolii, Bogdan Monogov, Alexander Potekhin, Maria Marina, Daniil Moskovskiy, Viktor Moskvoretskii, Alexander Panchenko, and Ilseyar Alimova. 2026. ToolHACE: A Dataset for Detecting Hallucinations in Tool-Based LLM Responses. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 40915–40934, Budapest, Hungary. Association for Computational Linguistics.