@inproceedings{mukhopadhyay-etal-2026-rubicon,
title = "{RUBICON}: Do Language Models Know When They{'}ve Crossed the Point of No Return?",
author = "Mukhopadhyay, Srija and
Qidwai, Adnan and
Sen, Jaydeep and
Shrivastava, Manish",
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.2704/",
pages = "61499--61530",
ISBN = "979-8-89176-459-0",
abstract = "As language models are increasingly deployed as autonomous agents, understanding whether they can anticipate the consequences of their actions is critical. We introduce RUBICON, a benchmark that tests whether models recognize when their own actions have made a task impossible to complete. Three procedurally generated puzzle domains measure this ability across spatial, resource management, and temporal reasoning, with a symbolic oracle providing ground truth. Evaluating eleven models, we find that all but one almost never flag a doomed situation on their own (0{--}7{\%} detection vs. 83{\%} for GPT-5.4), even though they correctly identify the same situations up to 100{\%} of the time when asked directly. Models whose reasoning traces contain correct unsolvability analysis frequently override it to provide a move anyway: self-override rates reach 34{--}77{\%} for models with high detection ability. When asked to identify which past decision caused the failure, models fail to trace back to the responsible action, instead blaming the moment the problem became obvious rather than the earlier decision that caused it. These findings highlight a fundamental gap: the ability to detect danger does not translate into the ability to avoid it."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="mukhopadhyay-etal-2026-rubicon">
<titleInfo>
<title>RUBICON: Do Language Models Know When They’ve Crossed the Point of No Return?</title>
</titleInfo>
<name type="personal">
<namePart type="given">Srija</namePart>
<namePart type="family">Mukhopadhyay</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Adnan</namePart>
<namePart type="family">Qidwai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jaydeep</namePart>
<namePart type="family">Sen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Manish</namePart>
<namePart type="family">Shrivastava</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>As language models are increasingly deployed as autonomous agents, understanding whether they can anticipate the consequences of their actions is critical. We introduce RUBICON, a benchmark that tests whether models recognize when their own actions have made a task impossible to complete. Three procedurally generated puzzle domains measure this ability across spatial, resource management, and temporal reasoning, with a symbolic oracle providing ground truth. Evaluating eleven models, we find that all but one almost never flag a doomed situation on their own (0–7% detection vs. 83% for GPT-5.4), even though they correctly identify the same situations up to 100% of the time when asked directly. Models whose reasoning traces contain correct unsolvability analysis frequently override it to provide a move anyway: self-override rates reach 34–77% for models with high detection ability. When asked to identify which past decision caused the failure, models fail to trace back to the responsible action, instead blaming the moment the problem became obvious rather than the earlier decision that caused it. These findings highlight a fundamental gap: the ability to detect danger does not translate into the ability to avoid it.</abstract>
<identifier type="citekey">mukhopadhyay-etal-2026-rubicon</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.2704/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>61499</start>
<end>61530</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T RUBICON: Do Language Models Know When They’ve Crossed the Point of No Return?
%A Mukhopadhyay, Srija
%A Qidwai, Adnan
%A Sen, Jaydeep
%A Shrivastava, Manish
%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 mukhopadhyay-etal-2026-rubicon
%X As language models are increasingly deployed as autonomous agents, understanding whether they can anticipate the consequences of their actions is critical. We introduce RUBICON, a benchmark that tests whether models recognize when their own actions have made a task impossible to complete. Three procedurally generated puzzle domains measure this ability across spatial, resource management, and temporal reasoning, with a symbolic oracle providing ground truth. Evaluating eleven models, we find that all but one almost never flag a doomed situation on their own (0–7% detection vs. 83% for GPT-5.4), even though they correctly identify the same situations up to 100% of the time when asked directly. Models whose reasoning traces contain correct unsolvability analysis frequently override it to provide a move anyway: self-override rates reach 34–77% for models with high detection ability. When asked to identify which past decision caused the failure, models fail to trace back to the responsible action, instead blaming the moment the problem became obvious rather than the earlier decision that caused it. These findings highlight a fundamental gap: the ability to detect danger does not translate into the ability to avoid it.
%U https://aclanthology.org/2026.emnlp-main.2704/
%P 61499-61530
Markdown (Informal)
[RUBICON: Do Language Models Know When They’ve Crossed the Point of No Return?](https://aclanthology.org/2026.emnlp-main.2704/) (Mukhopadhyay et al., EMNLP 2026)
ACL