Newest papers

Conflicted Records: Typed Adversarial Contracts for Auditable LLM Output Quality Enforcement

Kardelen Subasi, Jarosław A. Chudziak · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.29669 · cited by 0 works (OpenAlex) · open-access PDF

Portuguese

Os registros reproduzidos pelo LLM são registros operacionais estereogenerados, cada vez mais povoados por sistemas de informação de produção, onde eles são tratados como ativos de dados sujeitos às mesmas expectativas de qualidade que registros de bancos de dados operacionais. Um único registro pode simultaneamente satisfazer e violar uma regra de qualidade: um fato fundamentado adjacente a uma alucinação, uma resposta útil adjacente a um estereótipo. A limpeza de dados clássica não pode resolver isso porque os dois julgamentos são incompatíveis ao longo do mesmo intervalo; um único juiz LLM não pode nem, uma vez que compartilha os preconceitos de treinamento que produziram o conflito. Intervenções existentes como engenharia prompt, cadeia de pensamento, e RLHF reduzem as taxas

Machine translation (Meta M2M100); the English original follows.

Abstract

LLM-generated records increasingly populate production information systems where they are treated as data assets subject to the same quality expectations as records from operational databases. A single such record can simultaneously satisfy and violate a quality rule: a grounded fact adjacent to a hallucination, a useful answer adjacent to a stereotype. Classical data cleaning cannot resolve this because the two judgments are incompatible over the same span; a single LLM judge cannot either, since it shares the training biases that produced the conflict. Existing interventions such as prompt engineering, chain-of-thought, and RLHF reduce surface error rates but provide no audit trail and cannot distinguish genuine remediation from blanket suppression. The question is whether a deterministic check over structured inter-agent records can catch the defects that referential checks alone miss. We present BiasGuard-D: two adversarial roles drawn from distinct model families exchange typed claim documents, a three-model quorum adjudicates by majority vote, and a combined referential-plus-semantic invariant validates the verdict. The semantic component fires on verdicts referential checks miss; removing the iterative loop drops invariant satisfaction sharply. The deliberation layer transfers from stereotype to hallucination remediation on HaluEval QA with only role-framing changes, raising right-answer recovery by 49 percentage points; the triage gate required a cross-family second auditor to maintain recall on objective hallucination. Every claim, vote, and instruction is stored as a typed record giving operators a provenance trail auditable under emerging AI regulations.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Adversarial system, Audit trail, Computer security, Quality (philosophy), Audit

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange