@inproceedings{dzuyo-etal-2026-forensicbench,
title = "{F}orensic{B}ench: Evaluating Agentic {LLM}s on Journal-Entry Fraud Detection",
author = {Dzuyo, Guy St{\'e}phane Waffo and
Guibon, Ga{\"e}l and
Cerisara, Christophe and
Letelier, Luis Belmar},
editor = "Atanasova, Pepa and
Haffari, Gholamreza and
Wang, Minghan",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: Industry Track",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-industry.147/",
pages = "2473--2488",
ISBN = "979-8-89176-460-6",
abstract = "Accounting fraud costs organizations billions annually, yet auditors rarely confront it as a single suspicious posting. Instead, they investigate coordinated *fraud schemes* such as fictitious vendor payments, ghost-employee payroll, and inflated revenues that span many journal entries, each individually routine, and only become visible when understood against the underlying accounting process. Existing AI systems either score isolated entries as anomalies or mine company-level disclosures; none evaluate whether a large language model (LLM) agent can query a live ledger, recover coordinated fraud patterns, and assign the correct scheme type. We release **ForensicBench**: the first benchmark for evaluating agentic LLMs on *scheme-level* journal-entry fraud detection. The dataset, **Forensic Ledger**, extends DataSynth with five injected fraud scheme types and scheme-level labels linking journal entries to coordinated instances. We evaluate twelve open-weight models under a single reference agent scaffold on a public leaderboard: among them, the best reaches only 34.7{\%} Entry-F1, and Type-F1 is up to 13 points lower. We publicly release the dataset, evaluation harness, and a reference agentic baseline"
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="dzuyo-etal-2026-forensicbench">
<titleInfo>
<title>ForensicBench: Evaluating Agentic LLMs on Journal-Entry Fraud Detection</title>
</titleInfo>
<name type="personal">
<namePart type="given">Guy</namePart>
<namePart type="given">Stéphane</namePart>
<namePart type="given">Waffo</namePart>
<namePart type="family">Dzuyo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gaël</namePart>
<namePart type="family">Guibon</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Christophe</namePart>
<namePart type="family">Cerisara</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Luis</namePart>
<namePart type="given">Belmar</namePart>
<namePart type="family">Letelier</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: Industry Track</title>
</titleInfo>
<name type="personal">
<namePart type="given">Pepa</namePart>
<namePart type="family">Atanasova</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gholamreza</namePart>
<namePart type="family">Haffari</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Minghan</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">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-460-6</identifier>
</relatedItem>
<abstract>Accounting fraud costs organizations billions annually, yet auditors rarely confront it as a single suspicious posting. Instead, they investigate coordinated *fraud schemes* such as fictitious vendor payments, ghost-employee payroll, and inflated revenues that span many journal entries, each individually routine, and only become visible when understood against the underlying accounting process. Existing AI systems either score isolated entries as anomalies or mine company-level disclosures; none evaluate whether a large language model (LLM) agent can query a live ledger, recover coordinated fraud patterns, and assign the correct scheme type. We release **ForensicBench**: the first benchmark for evaluating agentic LLMs on *scheme-level* journal-entry fraud detection. The dataset, **Forensic Ledger**, extends DataSynth with five injected fraud scheme types and scheme-level labels linking journal entries to coordinated instances. We evaluate twelve open-weight models under a single reference agent scaffold on a public leaderboard: among them, the best reaches only 34.7% Entry-F1, and Type-F1 is up to 13 points lower. We publicly release the dataset, evaluation harness, and a reference agentic baseline</abstract>
<identifier type="citekey">dzuyo-etal-2026-forensicbench</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-industry.147/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>2473</start>
<end>2488</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T ForensicBench: Evaluating Agentic LLMs on Journal-Entry Fraud Detection
%A Dzuyo, Guy Stéphane Waffo
%A Guibon, Gaël
%A Cerisara, Christophe
%A Letelier, Luis Belmar
%Y Atanasova, Pepa
%Y Haffari, Gholamreza
%Y Wang, Minghan
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: Industry Track
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-460-6
%F dzuyo-etal-2026-forensicbench
%X Accounting fraud costs organizations billions annually, yet auditors rarely confront it as a single suspicious posting. Instead, they investigate coordinated *fraud schemes* such as fictitious vendor payments, ghost-employee payroll, and inflated revenues that span many journal entries, each individually routine, and only become visible when understood against the underlying accounting process. Existing AI systems either score isolated entries as anomalies or mine company-level disclosures; none evaluate whether a large language model (LLM) agent can query a live ledger, recover coordinated fraud patterns, and assign the correct scheme type. We release **ForensicBench**: the first benchmark for evaluating agentic LLMs on *scheme-level* journal-entry fraud detection. The dataset, **Forensic Ledger**, extends DataSynth with five injected fraud scheme types and scheme-level labels linking journal entries to coordinated instances. We evaluate twelve open-weight models under a single reference agent scaffold on a public leaderboard: among them, the best reaches only 34.7% Entry-F1, and Type-F1 is up to 13 points lower. We publicly release the dataset, evaluation harness, and a reference agentic baseline
%U https://aclanthology.org/2026.emnlp-industry.147/
%P 2473-2488
Markdown (Informal)
[ForensicBench: Evaluating Agentic LLMs on Journal-Entry Fraud Detection](https://aclanthology.org/2026.emnlp-industry.147/) (Dzuyo et al., EMNLP 2026)
ACL