Metraipolis › archive › dc_5c48de4a04ee
MasDrift: Benchmarking Authorization Preservation Across Multi-Agent Architectures — arXiv:2608.07556
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.07556
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-19T06:49:19.975Z
- SHA-256 of the archived text
724b5a70f88cd4b69ff6ddead09744c45086a647e75f6edda04cf08a5b84d11b- Stable URL of this page
https://metraipolis.com/documento/dc_5c48de4a04ee
How to cite
MasDrift: Benchmarking Authorization Preservation Across Multi-Agent Architectures — arXiv:2608.07556 — arXiv (en), https://arxiv.org/abs/2608.07556 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_5c48de4a04ee (Metraipolis archive dc_5c48de4a04ee, sha256 724b5a70f88cd4b69ff6ddead09744c45086a647e75f6edda04cf08a5b84d11b)
BibTeX
@misc{dc5c48de4a04ee,
title = {MasDrift: Benchmarking Authorization Preservation Across Multi-Agent Architectures — arXiv:2608.07556},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.07556},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_5c48de4a04ee; sha256 724b5a70f88cd4b69ff6ddead09744c45086a647e75f6edda04cf08a5b84d11b},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.07556. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Multi-agent systems (MAS) decompose long-horizon tasks across supervisors and subagents, but delegated goals do not necessarily carry their original authorization boundaries. Existing safety benchmarks mainly study adversarial compromise, while work on constraint drift lacks controlled architecture-level evaluation. We introduce MasDrift, a benchmark of 600 benign productivity tasks across eight domains. Each task pairs required work with reserved actions. MasDrift compares single-agent, centralized, and decentralized coordination while varying hierarchy depth and peer width, measuring task completion and authorization preservation. Across generic multi-agent conditions, centralized hierarchies achieve 93.9--98.6% task completion versus 85.7--87.0% for peer networks, while unauthorized actions occur in 2.7--19.8% of tasks versus 0.6--0.8%, a gap that widens with hierarchy depth. We further compare two defenses that differ in where authorization evidence resides. One re-anchors every pending call to the original user request. The other carries an attenuated policy along the delegation chain. Re-anchoring reduces unauthorized actions in every model configuration we evaluate, at a cost of 1.6 points of pooled completion. Chain propagation blocks required work instead, forfeiting up to 36.3 points. A heterogeneous case study confirms that the failure follows from coordination rather than model strength. MasDrift exposes a centralization tradeoff and makes authorization preservation a measurable property of MAS design. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-11 in cs.MA). Categorie arXiv: Multiagent Systems (cs.MA) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2608.07556 — 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 2026-10-19T06:45:31.675Z dalla pagina https://arxiv.org/catchup/cs.MA/2026-08-11?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).