Metraipolis › archive › dc_7ce4feed2da6
Knowing the Form, Not the Function: Automatically Auditing Answer--Authority Decoupling in Legal Benchmarks — arXiv:2608.02621
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.02621
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-31T12:17:25.695Z
- SHA-256 of the archived text
cc6f40394ab7c447de4302dc1d49737f193e8fa80c59cccbf56b87e808b96761- Stable URL of this page
https://metraipolis.com/documento/dc_7ce4feed2da6
How to cite
Knowing the Form, Not the Function: Automatically Auditing Answer--Authority Decoupling in Legal Benchmarks — arXiv:2608.02621 — arXiv (en), https://arxiv.org/abs/2608.02621 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7ce4feed2da6 (Metraipolis archive dc_7ce4feed2da6, sha256 cc6f40394ab7c447de4302dc1d49737f193e8fa80c59cccbf56b87e808b96761)
BibTeX
@misc{dc7ce4feed2da6,
title = {Knowing the Form, Not the Function: Automatically Auditing Answer--Authority Decoupling in Legal Benchmarks — arXiv:2608.02621},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.02621},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7ce4feed2da6; sha256 cc6f40394ab7c447de4302dc1d49737f193e8fa80c59cccbf56b87e808b96761},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.02621. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Legal benchmarks typically score final answers even when models also state legal authority. We test whether answer correctness can serve as a proxy for authority grounding. Under ordinary reasoning prompts that did not request statutory citations, four LLMs spontaneously produced authority markers across 238 Taiwan bar-examination items. Because each item has a verified governing provision, we automatically audit answer correctness and authority grounding jointly. The two dimensions dissociate in both directions. In criminal law, 24.0--42.4\% of valid responses were answer-correct but missed the gold authority, while 15.2--21.7\% were answer-incorrect but cited it. A separate statutory-retrieval probe and a permissive citation-abstention intervention further show that answer and citation behavior can move separately at the output level. Because this mismatch arises without adversarial or inconsistency-inducing prompting, answer-only scoring treats naturally occurring gold-authority misses as complete benchmark successes. Because statutory authority is structurally extractable and externally verifiable, the failure can be measured automatically. A preliminary PRC civil-law extension also observes citation-unrequested authority marking, motivating a full cross-jurisdictional joint audit. We therefore propose joint answer--authority evaluation for statute-grounded legal benchmarks. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-05 in cs.AI). Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2608.02621 — 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 2026-10-31T12:05:52.892Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-05?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).