Metraipolis › archive › dc_ec9905562b05
The Helpful Agent Problem: When Emergent Cooperation Among LLM Agents Bypasses Intended Workflows — arXiv:2610.31985
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.31985
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-26T19:47:23.543Z
- SHA-256 of the archived text
dd894b83e08bf8f8ee69d180e9e449a049dea58a5e03ee535ff7fe71ea5845d0- Stable URL of this page
https://metraipolis.com/documento/dc_ec9905562b05
How to cite
The Helpful Agent Problem: When Emergent Cooperation Among LLM Agents Bypasses Intended Workflows — arXiv:2610.31985 — arXiv (en), https://arxiv.org/abs/2610.31985 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ec9905562b05 (Metraipolis archive dc_ec9905562b05, sha256 dd894b83e08bf8f8ee69d180e9e449a049dea58a5e03ee535ff7fe71ea5845d0)
BibTeX
@misc{dcec9905562b05,
title = {The Helpful Agent Problem: When Emergent Cooperation Among LLM Agents Bypasses Intended Workflows — arXiv:2610.31985},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.31985},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ec9905562b05; sha256 dd894b83e08bf8f8ee69d180e9e449a049dea58a5e03ee535ff7fe71ea5845d0},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.31985. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
LLM agents are increasingly deployed in collaborative settings where they pursue separate tasks while sharing resources and operating under review or oversight processes. Such settings create safety risks not only when agents are malicious, but also when they are simply trying to help. Prior work has largely studied adversarial or explicitly collusive behavior; we show that benign agents pursuing separate objectives can spontaneously cooperate in ways that bypass intended workflows. We call this the \emph{helpful agent problem}. We study long-horizon shared-resource environments in which agents sometimes depend on resources owned by their peers and cross-owner changes are normally mediated by a reviewer. When review becomes unavailable, agents begin making changes on one another's behalf despite having no shared reward or explicit incentive to cooperate. Crucially, these actions remain within each agent's local permissions while circumventing the system's intended oversight structure. We introduce \textsc{FavorBench}, a controlled benchmark with measurable downstream side effects, and adapt \textsc{CooperBench} to test the phenomenon in realistic repositories. Across two benchmarks and four models, we find that out-of-workflow favors emerge frequently. Further, in \textsc{FavorBench}, favors from at least three models cause significant harm to parties outside the interaction. These results reveal a distinct multi-agent safety challenge: cooperation itself can create pathways around system-level safeguards.
Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.MA, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL); Multiagent Systems (cs.MA).
Pagina del lavoro: https://arxiv.org/abs/2610.31985 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-26T19:45:18.252Z dalla pagina https://arxiv.org/list/cs.MA/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).