Metraipolis › archive › dc_a4564290fea8

When Debate Amplifies Risk: Social-Bias Vulnerabilities in Multi-Agent Safety Deliberation — arXiv:2611.04289

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.04289
License (as archived)
CC0-1.0
Archived
2026-11-04T21:39:56.781Z
SHA-256 of the archived text
3d8d94b04ce134a257f17f20ce8f673215e6036aa75ac59efa9144f3b9a94b5c
Stable URL of this page
https://metraipolis.com/documento/dc_a4564290fea8

How to cite

When Debate Amplifies Risk: Social-Bias Vulnerabilities in Multi-Agent Safety Deliberation — arXiv:2611.04289 — arXiv (en), https://arxiv.org/abs/2611.04289 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_a4564290fea8 (Metraipolis archive dc_a4564290fea8, sha256 3d8d94b04ce134a257f17f20ce8f673215e6036aa75ac59efa9144f3b9a94b5c)
BibTeX
@misc{dca4564290fea8,
  title = {When Debate Amplifies Risk: Social-Bias Vulnerabilities in Multi-Agent Safety Deliberation — arXiv:2611.04289},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.04289},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_a4564290fea8; sha256 3d8d94b04ce134a257f17f20ce8f673215e6036aa75ac59efa9144f3b9a94b5c},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.04289. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Retrieval-augmented Web applications increasingly employ multi-agent debate (MAD) to aggregate conflicting, noisy, and heterogeneous information, yet deliberation is fundamentally a social process susceptible to interpersonal dynamics. When retrieved Web sources contain latent hazards alongside safe factual content, social biases such as conformity pressure and authority deference can override evidence-grounded safety boundaries. In this work, we investigate whether the social structure of deliberation can paradoxically amplify latent Web safety risks. We introduce SafeConflict-Bias, a controlled counterfactual diagnostic protocol spanning six core Web risk categories, which holds queries and retrieved evidence strictly invariant while isolating conversational majority sway, authority cues, and confidence asymmetry. Empirical evaluation across diverse proprietary and open-source backbones reveals that while isolated models reliably reject unsafe guidance, standard MAD catastrophically capitulates to social pressure---surging unsafe adoption from 0.0\% up to 100.0\% and systematically silencing safe minority evidence. To resolve this systemic fragility without retraining or debate regeneration, we propose RiskTrace-Debias, an inference-time auditing framework uniting evidence-first precommitment, social-metadata masking, and consistency-gated arbitration. RiskTrace-Debias suppresses worst-case unsafe adoption to $\le 2.0\%$ across all architectures without inducing over-refusal, while generating verifiable provenance traces to audit interpersonal manipulation in Web deliberation.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Cryptography and Security (cs.CR) ; Artificial Intelligence (cs.AI); Computers and Society (cs.CY).
Pagina del lavoro: https://arxiv.org/abs/2611.04289 — 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 2026-11-04T21:35:49.658Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).