@inproceedings{goel-gurevych-2026-capable,
title = "Capable but Careless: Do Computer-Use Agents Follow Contextual Integrity?",
author = "Goel, Anmol and
Gurevych, Iryna",
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.55/",
pages = "1235--1266",
ISBN = "979-8-89176-459-0",
abstract = "Computer-use agents (CUAs) now act on a user{'}s behalf across personal applications such as email, calendars, and to-do lists. This cross-application access is useful, but it also creates a privacy risk that has been largely overlooked: when an agent works in one context, it can pull in information from another that is inappropriate in that context. Hence, we introduce , a benchmark that turns this risk into executable, deterministically scored scenarios. We target three common failure modes in CUAs: \textit{visual co-location}, where the agent pulls in prohibited items that sit next to the task target in the UI; \textit{task-ambiguity overshare}, where the agent dumps dense personal state in response to an under-specified prompt; and \textit{recipient misalignment}, where the agent sends content to an addressee for whom it is inappropriate. We evaluate 15 frontier agents and find a surprisingly high failure rate: 11 of 15 leak on more than 50{\%} of scenarios, with an average leakage of 67.9{\%}, and the same failures persist when agents act end-to-end in the environment to complete the task. We release to encourage the development of safer computer-use agents and position contextual disclosure testing as a pre-deployment safety check."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="goel-gurevych-2026-capable">
<titleInfo>
<title>Capable but Careless: Do Computer-Use Agents Follow Contextual Integrity?</title>
</titleInfo>
<name type="personal">
<namePart type="given">Anmol</namePart>
<namePart type="family">Goel</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Iryna</namePart>
<namePart type="family">Gurevych</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>Computer-use agents (CUAs) now act on a user’s behalf across personal applications such as email, calendars, and to-do lists. This cross-application access is useful, but it also creates a privacy risk that has been largely overlooked: when an agent works in one context, it can pull in information from another that is inappropriate in that context. Hence, we introduce , a benchmark that turns this risk into executable, deterministically scored scenarios. We target three common failure modes in CUAs: visual co-location, where the agent pulls in prohibited items that sit next to the task target in the UI; task-ambiguity overshare, where the agent dumps dense personal state in response to an under-specified prompt; and recipient misalignment, where the agent sends content to an addressee for whom it is inappropriate. We evaluate 15 frontier agents and find a surprisingly high failure rate: 11 of 15 leak on more than 50% of scenarios, with an average leakage of 67.9%, and the same failures persist when agents act end-to-end in the environment to complete the task. We release to encourage the development of safer computer-use agents and position contextual disclosure testing as a pre-deployment safety check.</abstract>
<identifier type="citekey">goel-gurevych-2026-capable</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.55/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>1235</start>
<end>1266</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Capable but Careless: Do Computer-Use Agents Follow Contextual Integrity?
%A Goel, Anmol
%A Gurevych, Iryna
%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 goel-gurevych-2026-capable
%X Computer-use agents (CUAs) now act on a user’s behalf across personal applications such as email, calendars, and to-do lists. This cross-application access is useful, but it also creates a privacy risk that has been largely overlooked: when an agent works in one context, it can pull in information from another that is inappropriate in that context. Hence, we introduce , a benchmark that turns this risk into executable, deterministically scored scenarios. We target three common failure modes in CUAs: visual co-location, where the agent pulls in prohibited items that sit next to the task target in the UI; task-ambiguity overshare, where the agent dumps dense personal state in response to an under-specified prompt; and recipient misalignment, where the agent sends content to an addressee for whom it is inappropriate. We evaluate 15 frontier agents and find a surprisingly high failure rate: 11 of 15 leak on more than 50% of scenarios, with an average leakage of 67.9%, and the same failures persist when agents act end-to-end in the environment to complete the task. We release to encourage the development of safer computer-use agents and position contextual disclosure testing as a pre-deployment safety check.
%U https://aclanthology.org/2026.emnlp-main.55/
%P 1235-1266
Markdown (Informal)
[Capable but Careless: Do Computer-Use Agents Follow Contextual Integrity?](https://aclanthology.org/2026.emnlp-main.55/) (Goel & Gurevych, EMNLP 2026)
ACL