Metraipolis › archive › dc_494223fa394c

Four Bridges: Measuring Deception Propensity in a Multi-Agent Game — arXiv:2611.03600

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.03600
License (as archived)
CC0-1.0
Archived
2026-11-03T18:09:52.647Z
SHA-256 of the archived text
7c465c23a6a5ef939bdf3b69df781cf5e699d67ab110e5cae371ed96bc498a8a
Stable URL of this page
https://metraipolis.com/documento/dc_494223fa394c

How to cite

Four Bridges: Measuring Deception Propensity in a Multi-Agent Game — arXiv:2611.03600 — arXiv (en), https://arxiv.org/abs/2611.03600 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_494223fa394c (Metraipolis archive dc_494223fa394c, sha256 7c465c23a6a5ef939bdf3b69df781cf5e699d67ab110e5cae371ed96bc498a8a)
BibTeX
@misc{dc494223fa394c,
  title = {Four Bridges: Measuring Deception Propensity in a Multi-Agent Game — arXiv:2611.03600},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.03600},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_494223fa394c; sha256 7c465c23a6a5ef939bdf3b69df781cf5e699d67ab110e5cae371ed96bc498a8a},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.03600. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

AI agents are increasingly deployed in settings where they interact and coordinate with other agents, yet honesty is mostly measured in single-turn, single-agent benchmarks that pressure the model to deceive. We study deception that emerges in multi-agent interaction without any instruction to deceive. We introduce Kradle, a platform for evaluating AI models, and Four Bridges, an information asymmetry game in which four agents distribute across four rooms: one is the death room, the others hold food worth more if not shared. One agent, the informed agent, knows which room is the death room and can withhold or disclose that information. We evaluate eight frontier models as the informed agent. Deceptive behavior appears in every model and in 73% of all runs (6% to 100% per model), and does not decrease with capability tier within a model family. When the informed agent does disclose, the warning is often challenged, and agents that challenge it enter the death room twice as often as those that do not. Telling all agents that one of them knows the death room collapses deception in six of eight models; telling the informed agent that the room is permanently fatal moves only one. On an open-weight model, an honesty steering vector reduces deception to 0% and adversarial honesty training to 6%. The platform and game provide a testbed for measuring deceptive behavior and comparing interventions.

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/2611.03600 — 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 2026-11-03T18:09:37.701Z dalla pagina https://arxiv.org/list/cs.MA/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).