Metraipolis › archive › dc_e1cd61848165
DelegationBench: Measuring When AI Agents Should Ask Before Acting — arXiv:2610.05532
- Source
- arXiv
- Original
- https://arxiv.org/abs/2610.05532
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-06T05:25:14.379Z
- SHA-256 of the archived text
7baab8bec249212b48ec0673e89b6547b89f27415737674564276f84b18425a2- Stable URL of this page
https://metraipolis.com/documento/dc_e1cd61848165
How to cite
DelegationBench: Measuring When AI Agents Should Ask Before Acting — arXiv:2610.05532 — arXiv, https://arxiv.org/abs/2610.05532 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e1cd61848165 (Metraipolis archive dc_e1cd61848165, sha256 7baab8bec249212b48ec0673e89b6547b89f27415737674564276f84b18425a2)
BibTeX
@misc{dce1cd61848165,
title = {DelegationBench: Measuring When AI Agents Should Ask Before Acting — arXiv:2610.05532},
howpublished = {arXiv},
url = {https://arxiv.org/abs/2610.05532},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e1cd61848165; sha256 7baab8bec249212b48ec0673e89b6547b89f27415737674564276f84b18425a2},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.05532. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
AI agents that send emails, edit files, and make purchases must decide when to act on their own and when to check with the user first. This decision is usually evaluated by showing a model a proposed action, asking whether it should proceed, and scoring agreement with human labels. We introduce DelegationBench to test whether such scores can be trusted. It has 156 scenarios with four possible responses (act, ask for permission, ask for missing information, refuse), and most scenarios come in matched pairs that change a single feature: whether the action was requested, what is at stake, whether it can be undone, or who will see it. Across ten models from five families, agreement scores mislead in three ways. A simple keyword rule, which we wrote after seeing the benchmark, agrees with our annotators more often than eight of the models, yet its decision changes in only 9 of 48 matched pairs. Equivalent ways of asking the same question change how often a model acts by up to 52.5 percentage points. And every model stops to ask the user less often when it must carry out the task with tools than when it judges a proposed action. When rules are stated explicitly, the same models follow them almost perfectly, so the gaps are not explained by a general inability to follow rules. We release the benchmark and evaluation tools and recommend reporting these properties separately rather than as one score. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL); Human-Computer Interaction (cs.HC); Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2610.05532 — 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 2026-10-06T05:24:05.604Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).