@inproceedings{jang-etal-2026-evaluating,
title = "Evaluating Coding Agents on Kernel Exploit Generation",
author = "Jang, Junyoung and
Lee, Gwanhyun and
Lee, Hwiwon and
Kim, Kyuheon and
Kim, Jongseong and
Jung, Jinho and
Zhang, Lingming",
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.1653/",
pages = "37581--37603",
ISBN = "979-8-89176-459-0",
abstract = "Coding agents now find real vulnerabilities in production software. However, bug discovery results do not measure whether agents can construct exploit primitives. We introduce KEX-bench, a benchmark for evaluating coding agents on exploit primitive generation against real operating-system kernels. KEX-bench contains 45 task instances across 40 Linux and Windows CVEs, covering kernel address leak, instruction-pointer control, heap read, heap write, and arbitrary address write. Each task runs in an isolated virtual machine, exposes controlled tools, and uses a deterministic verifier for checking primitive-specific success. We evaluate state-of-the-art coding agents paired with frontier and open-weight models under fixed tool-call budgets. Without a PoC, the strongest configuration solves 1 of 20 Windows tasks (5.0{\%}) and 14 of 25 Linux tasks (56.0{\%}). With a reference PoC, the strongest configuration solves 31 of 45 tasks (68.9{\%}). This highlights the gap where agents reach kernel crashes but fail to shape kernel state into exploit primitives. We release KEX-bench for reproducible research on AI-assisted exploitation at https://kex-bench.github.io."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="jang-etal-2026-evaluating">
<titleInfo>
<title>Evaluating Coding Agents on Kernel Exploit Generation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Junyoung</namePart>
<namePart type="family">Jang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gwanhyun</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hwiwon</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kyuheon</namePart>
<namePart type="family">Kim</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jongseong</namePart>
<namePart type="family">Kim</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jinho</namePart>
<namePart type="family">Jung</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lingming</namePart>
<namePart type="family">Zhang</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>Coding agents now find real vulnerabilities in production software. However, bug discovery results do not measure whether agents can construct exploit primitives. We introduce KEX-bench, a benchmark for evaluating coding agents on exploit primitive generation against real operating-system kernels. KEX-bench contains 45 task instances across 40 Linux and Windows CVEs, covering kernel address leak, instruction-pointer control, heap read, heap write, and arbitrary address write. Each task runs in an isolated virtual machine, exposes controlled tools, and uses a deterministic verifier for checking primitive-specific success. We evaluate state-of-the-art coding agents paired with frontier and open-weight models under fixed tool-call budgets. Without a PoC, the strongest configuration solves 1 of 20 Windows tasks (5.0%) and 14 of 25 Linux tasks (56.0%). With a reference PoC, the strongest configuration solves 31 of 45 tasks (68.9%). This highlights the gap where agents reach kernel crashes but fail to shape kernel state into exploit primitives. We release KEX-bench for reproducible research on AI-assisted exploitation at https://kex-bench.github.io.</abstract>
<identifier type="citekey">jang-etal-2026-evaluating</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1653/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>37581</start>
<end>37603</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Evaluating Coding Agents on Kernel Exploit Generation
%A Jang, Junyoung
%A Lee, Gwanhyun
%A Lee, Hwiwon
%A Kim, Kyuheon
%A Kim, Jongseong
%A Jung, Jinho
%A Zhang, Lingming
%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 jang-etal-2026-evaluating
%X Coding agents now find real vulnerabilities in production software. However, bug discovery results do not measure whether agents can construct exploit primitives. We introduce KEX-bench, a benchmark for evaluating coding agents on exploit primitive generation against real operating-system kernels. KEX-bench contains 45 task instances across 40 Linux and Windows CVEs, covering kernel address leak, instruction-pointer control, heap read, heap write, and arbitrary address write. Each task runs in an isolated virtual machine, exposes controlled tools, and uses a deterministic verifier for checking primitive-specific success. We evaluate state-of-the-art coding agents paired with frontier and open-weight models under fixed tool-call budgets. Without a PoC, the strongest configuration solves 1 of 20 Windows tasks (5.0%) and 14 of 25 Linux tasks (56.0%). With a reference PoC, the strongest configuration solves 31 of 45 tasks (68.9%). This highlights the gap where agents reach kernel crashes but fail to shape kernel state into exploit primitives. We release KEX-bench for reproducible research on AI-assisted exploitation at https://kex-bench.github.io.
%U https://aclanthology.org/2026.emnlp-main.1653/
%P 37581-37603
Markdown (Informal)
[Evaluating Coding Agents on Kernel Exploit Generation](https://aclanthology.org/2026.emnlp-main.1653/) (Jang et al., EMNLP 2026)
ACL
- Junyoung Jang, Gwanhyun Lee, Hwiwon Lee, Kyuheon Kim, Jongseong Kim, Jinho Jung, and Lingming Zhang. 2026. Evaluating Coding Agents on Kernel Exploit Generation. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 37581–37603, Budapest, Hungary. Association for Computational Linguistics.