Metraipolis › archive › dc_f1a7ecd9d4eb

Learning to Report Unsafe Tasks in a Multi-Agent Game — arXiv:2610.09002

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.09002
License (as archived)
CC0-1.0
Archived
2026-10-08T11:02:15.036Z
SHA-256 of the archived text
62da228e085440ce13703c2198af24df677cfdd108221a4c4a47dde63a82386a
Stable URL of this page
https://metraipolis.com/documento/dc_f1a7ecd9d4eb

How to cite

Learning to Report Unsafe Tasks in a Multi-Agent Game — arXiv:2610.09002 — arXiv (en), https://arxiv.org/abs/2610.09002 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f1a7ecd9d4eb (Metraipolis archive dc_f1a7ecd9d4eb, sha256 62da228e085440ce13703c2198af24df677cfdd108221a4c4a47dde63a82386a)
BibTeX
@misc{dcf1a7ecd9d4eb,
  title = {Learning to Report Unsafe Tasks in a Multi-Agent Game — arXiv:2610.09002},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.09002},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f1a7ecd9d4eb; sha256 62da228e085440ce13703c2198af24df677cfdd108221a4c4a47dde63a82386a},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.09002. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

When agents share a reward for completed tasks, reporting unsafe work can reduce the reporter's reward by stopping a task. Audits can make reporting optimal without ensuring that further training teaches a silent team to report. We study this learning problem in a game where any witness can stop a task by reporting. With $k$ witnesses per task sharing a policy and drawing independently, the expected-reward derivative with respect to their shared silence probability counts each task's benefit $k$ times at universal silence. The comparison with universal reporting counts it once. For arbitrary policy groups, we give an audit condition sufficient for exact policy-gradient updates to reach universal reporting and, apart from boundary cases, necessary near universal silence. In a balanced family, the cheapest audits meeting the condition with prescribed positive margins cost exactly $k$ times as much for full sharing as for one policy per role. We train PPO policies on 24 witness graphs from learned silence. Separating co-witnesses reduces unsafe completion by 33.59 percentage points compared with shuffled groups of the same sizes under the same audits (95% graph-bootstrap interval: 21.03-45.13). Only 9 of 48 witness-group runs achieve below 1% unsafe completion while retaining at least 90% legitimate completion. At the same audit budget, a fully shared network meets both thresholds in none of 48 runs with independent action draws and all 48 with a common draw.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.MA, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Multiagent Systems (cs.MA).
Pagina del lavoro: https://arxiv.org/abs/2610.09002 — 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 2026-10-08T10:09:05.148Z dalla pagina https://arxiv.org/list/cs.MA/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).