<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="barbi-etal-2025-preventing">
    <titleInfo>
        <title>Preventing Rogue Agents Improves Multi-Agent Collaboration</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Ohav</namePart>
        <namePart type="family">Barbi</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Ori</namePart>
        <namePart type="family">Yoran</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Mor</namePart>
        <namePart type="family">Geva</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2025-07</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the 1st Workshop for Research on Agent Language Models (REALM 2025)</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Ehsan</namePart>
            <namePart type="family">Kamalloo</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Nicolas</namePart>
            <namePart type="family">Gontier</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Xing</namePart>
            <namePart type="given">Han</namePart>
            <namePart type="family">Lu</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Nouha</namePart>
            <namePart type="family">Dziri</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Shikhar</namePart>
            <namePart type="family">Murty</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Alexandre</namePart>
            <namePart type="family">Lacoste</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Vienna, Austria</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-264-0</identifier>
    </relatedItem>
    <abstract>Multi-agent systems, where specialized agents collaborate to solve a shared task hold great potential, from increased modularity to simulating complex environments. However, they also have a major caveat – a single agent can cause the entire system to fail. Consider a simple game where the knowledge to solve the task is distributed between agents, which share information in a communication channel. At each round, any of the agents can terminate the game and make the final prediction, even if they are uncertain about the outcome of their action. Detection of such rogue agents __before they act__ may prevent the system’s failure. In this work, we propose to __monitor__ agents during action prediction and __intervene__ when a future error is likely to occur. To test our approach, we introduce WhoDunitEnv, a multi-agent collaboration environment that allows modular control over task complexity and communication structure. Experiments on WhoDunitEnv, code generation tasks and the GovSim environment for resource sustainability show that our approach leads to substantial performance gains up to 17.4%, 2.5% and 20%, respectively. Thorough analysis shows that our monitors successfully identify critical points of agent confusion and our interventions effectively stop agent errors from propagating.</abstract>
    <identifier type="citekey">barbi-etal-2025-preventing</identifier>
    <location>
        <url>https://aclanthology.org/2025.realm-1.34/</url>
    </location>
    <part>
        <date>2025-07</date>
        <extent unit="page">
            <start>486</start>
            <end>511</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Preventing Rogue Agents Improves Multi-Agent Collaboration
%A Barbi, Ohav
%A Yoran, Ori
%A Geva, Mor
%Y Kamalloo, Ehsan
%Y Gontier, Nicolas
%Y Lu, Xing Han
%Y Dziri, Nouha
%Y Murty, Shikhar
%Y Lacoste, Alexandre
%S Proceedings of the 1st Workshop for Research on Agent Language Models (REALM 2025)
%D 2025
%8 July
%I Association for Computational Linguistics
%C Vienna, Austria
%@ 979-8-89176-264-0
%F barbi-etal-2025-preventing
%X Multi-agent systems, where specialized agents collaborate to solve a shared task hold great potential, from increased modularity to simulating complex environments. However, they also have a major caveat – a single agent can cause the entire system to fail. Consider a simple game where the knowledge to solve the task is distributed between agents, which share information in a communication channel. At each round, any of the agents can terminate the game and make the final prediction, even if they are uncertain about the outcome of their action. Detection of such rogue agents __before they act__ may prevent the system’s failure. In this work, we propose to __monitor__ agents during action prediction and __intervene__ when a future error is likely to occur. To test our approach, we introduce WhoDunitEnv, a multi-agent collaboration environment that allows modular control over task complexity and communication structure. Experiments on WhoDunitEnv, code generation tasks and the GovSim environment for resource sustainability show that our approach leads to substantial performance gains up to 17.4%, 2.5% and 20%, respectively. Thorough analysis shows that our monitors successfully identify critical points of agent confusion and our interventions effectively stop agent errors from propagating.
%U https://aclanthology.org/2025.realm-1.34/
%P 486-511

Markdown (Informal)

[Preventing Rogue Agents Improves Multi-Agent Collaboration](https://aclanthology.org/2025.realm-1.34/) (Barbi et al., REALM 2025)

• Preventing Rogue Agents Improves Multi-Agent Collaboration (Barbi et al., REALM 2025)

ACL

• Ohav Barbi, Ori Yoran, and Mor Geva. 2025. Preventing Rogue Agents Improves Multi-Agent Collaboration. In Proceedings of the 1st Workshop for Research on Agent Language Models (REALM 2025), pages 486–511, Vienna, Austria. Association for Computational Linguistics.