@inproceedings{kim-etal-2026-carob,
title = "{CAROB}: A Benchmark for Cross-Artifact and Authority-aware Reasoning over Organizational Bundles",
author = "Kim, Eunbin and
Namgoong, Hyuk and
Go, Wooyoung and
Ryu, Seungjin and
Jung, Sangkeun",
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.2185/",
pages = "46786--46804",
ISBN = "979-8-89176-462-0",
abstract = "Enterprise AI assistants suffer from two failure modes not captured by existing public-web QA benchmarks: authorized information leakage, where factually correct answers reach unintended readers, and aggregation failure, where evidence scattered across artifacts yields fluent but incomplete responses. Both stem from enterprise authority being per-item and partly body-embedded, a structure that existing benchmarks address along at most one axis. We argue that authority-aware evaluation must be conducted at the per-item level, and instantiate this principle through Cross-Artifact and Authority-aware Reasoning over Organizational Bundles (CAROB): a synthetic enterprise dataset in which every retrieval-and-citation item carries a time-indexed audience/approval/lifecycle annotation, paired with three tasks that decouple cross-artifact coherence, authority-aware retrieval, and role-conditioned answering. Evaluating five frontier and open-weight LLMs, we find that no single model dominates both axes of the benchmark: each task induces a distinct model ordering, and the model strongest on authority handling ranks last on answer coverage. Two structural patterns hold across the lineup: retrieval-side leakage amplifies substantially at the answer stage, and structural authority filtering eliminates the residual leak band that prompt-level mitigation leaves behind, with both patterns replicating on EnronQA and ACCESS DENIED INC. We release CAROB as an evaluation testbed for authority-aware enterprise system design, together with its per-item annotation, three-task structure, and label-first generation pipeline as reusable infrastructure."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="kim-etal-2026-carob">
<titleInfo>
<title>CAROB: A Benchmark for Cross-Artifact and Authority-aware Reasoning over Organizational Bundles</title>
</titleInfo>
<name type="personal">
<namePart type="given">Eunbin</namePart>
<namePart type="family">Kim</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hyuk</namePart>
<namePart type="family">Namgoong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wooyoung</namePart>
<namePart type="family">Go</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Seungjin</namePart>
<namePart type="family">Ryu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sangkeun</namePart>
<namePart type="family">Jung</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>Enterprise AI assistants suffer from two failure modes not captured by existing public-web QA benchmarks: authorized information leakage, where factually correct answers reach unintended readers, and aggregation failure, where evidence scattered across artifacts yields fluent but incomplete responses. Both stem from enterprise authority being per-item and partly body-embedded, a structure that existing benchmarks address along at most one axis. We argue that authority-aware evaluation must be conducted at the per-item level, and instantiate this principle through Cross-Artifact and Authority-aware Reasoning over Organizational Bundles (CAROB): a synthetic enterprise dataset in which every retrieval-and-citation item carries a time-indexed audience/approval/lifecycle annotation, paired with three tasks that decouple cross-artifact coherence, authority-aware retrieval, and role-conditioned answering. Evaluating five frontier and open-weight LLMs, we find that no single model dominates both axes of the benchmark: each task induces a distinct model ordering, and the model strongest on authority handling ranks last on answer coverage. Two structural patterns hold across the lineup: retrieval-side leakage amplifies substantially at the answer stage, and structural authority filtering eliminates the residual leak band that prompt-level mitigation leaves behind, with both patterns replicating on EnronQA and ACCESS DENIED INC. We release CAROB as an evaluation testbed for authority-aware enterprise system design, together with its per-item annotation, three-task structure, and label-first generation pipeline as reusable infrastructure.</abstract>
<identifier type="citekey">kim-etal-2026-carob</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.2185/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>46786</start>
<end>46804</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T CAROB: A Benchmark for Cross-Artifact and Authority-aware Reasoning over Organizational Bundles
%A Kim, Eunbin
%A Namgoong, Hyuk
%A Go, Wooyoung
%A Ryu, Seungjin
%A Jung, Sangkeun
%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 kim-etal-2026-carob
%X Enterprise AI assistants suffer from two failure modes not captured by existing public-web QA benchmarks: authorized information leakage, where factually correct answers reach unintended readers, and aggregation failure, where evidence scattered across artifacts yields fluent but incomplete responses. Both stem from enterprise authority being per-item and partly body-embedded, a structure that existing benchmarks address along at most one axis. We argue that authority-aware evaluation must be conducted at the per-item level, and instantiate this principle through Cross-Artifact and Authority-aware Reasoning over Organizational Bundles (CAROB): a synthetic enterprise dataset in which every retrieval-and-citation item carries a time-indexed audience/approval/lifecycle annotation, paired with three tasks that decouple cross-artifact coherence, authority-aware retrieval, and role-conditioned answering. Evaluating five frontier and open-weight LLMs, we find that no single model dominates both axes of the benchmark: each task induces a distinct model ordering, and the model strongest on authority handling ranks last on answer coverage. Two structural patterns hold across the lineup: retrieval-side leakage amplifies substantially at the answer stage, and structural authority filtering eliminates the residual leak band that prompt-level mitigation leaves behind, with both patterns replicating on EnronQA and ACCESS DENIED INC. We release CAROB as an evaluation testbed for authority-aware enterprise system design, together with its per-item annotation, three-task structure, and label-first generation pipeline as reusable infrastructure.
%U https://aclanthology.org/2026.findings-emnlp.2185/
%P 46786-46804
Markdown (Informal)
[CAROB: A Benchmark for Cross-Artifact and Authority-aware Reasoning over Organizational Bundles](https://aclanthology.org/2026.findings-emnlp.2185/) (Kim et al., Findings 2026)
ACL