Metraipolis › archive › dc_eefab8d0ad16

Do LLMs Trust the Accuser or the Accusation? Measuring Belief Shifts in Werewolf — arXiv:2609.12446

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.12446
License (as archived)
CC0-1.0
Archived
2026-10-16T03:02:45.694Z
SHA-256 of the archived text
9ef284aed589373efc548c24f29ce55e11bb97edcbde9e0e8c4ba4d7b1af22cf
Stable URL of this page
https://metraipolis.com/documento/dc_eefab8d0ad16

How to cite

Do LLMs Trust the Accuser or the Accusation? Measuring Belief Shifts in Werewolf — arXiv:2609.12446 — arXiv (en), https://arxiv.org/abs/2609.12446 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_eefab8d0ad16 (Metraipolis archive dc_eefab8d0ad16, sha256 9ef284aed589373efc548c24f29ce55e11bb97edcbde9e0e8c4ba4d7b1af22cf)
BibTeX
@misc{dceefab8d0ad16,
  title = {Do LLMs Trust the Accuser or the Accusation? Measuring Belief Shifts in Werewolf — arXiv:2609.12446},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.12446},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_eefab8d0ad16; sha256 9ef284aed589373efc548c24f29ce55e11bb97edcbde9e0e8c4ba4d7b1af22cf},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.12446. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Social-deduction games such as Werewolf are increasingly used to evaluate LLM agents, but existing evaluations often rely on final game outcomes. We propose a belief-shift evaluation benchmark in Werewolf for analyzing communication skills through belief updating. Using LLM-played games, we annotate suspicion and accusation messages and measure how an observing village-side model's beliefs change after each message. We evaluate 40 open-weight LLM configurations on 1,224 annotated messages. Our results show that larger models better distinguish true wolves from villagers based on game history, but accusations still strongly influence their beliefs. Models become more suspicious of the accused target and less suspicious of the accuser, especially when the accuser is trusted, even if the accuser is wolf-aligned. Larger models better resist accusations from accusers they already distrust. Overall, our findings suggest that current open-weight LLMs up to 120B parameters still struggle to integrate accusation content with source trust in strategic communication. Our benchmark and code are available at this https URL .

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-14 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.12446 — 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 2026-10-16T02:51:27.794Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-14?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).