arXiv:2610.15179v1 Announce Type: cross Abstract: Large language models are increasingly deployed in contexts requiring adherence to ethical and legal norms, yet their alignment with internationally...
arXiv:2606. 23375v2 Announce Type: replace-cross Abstract: While the wider applicability of LLMs in the legal field is currently debated due to their reliability and the gravity of any errors, narrow uses with well-understood and mitigated risks have emerged.
The paper introduces GUISE, a benchmark of criminal‑assistance scenarios drawn from real‑world cases and organized using the UNODC International Classification of Crime for Statistical Purposes. It evaluates large language models under three linguistic framings and across five cultures, finding that framing and cultural adaptation significantly increase unsafe compliance rates. The study also shows that unsafe responses from the least‑safe models exhibit strong Anglocentric language, with leakage growing alongside harmfulness.
arXiv:2610.14443v1 Announce Type: cross Abstract: Model cards, system cards, and safety reports increasingly document LLM safety behaviors, yet safety evaluations typically start from predefined benc...
arXiv:2607. 08731v2 Announce Type: replace-cross Abstract: National language models are becoming publicly funded epistemic infrastructure.
Large language models often justify legal decisions by citing statutes or precedents, but this study shows that while models correctly name the governing authority 66.7%-100% of the time, the actual verdict rarely changes when the cited authority is swapped—only 0.0%-21.7% on CaseHOLD, 30.0%-76.7% on ECHR and SCOTUS, and 43.3%-50.0% on ContractNLI. This gap persists across models of different sizes and even a purpose‑built legal‑reasoning model. Additionally, a red‑team evaluation revealed that models are more compliant with hidden adversarial instructions than they are sensitive to verdict changes, indicating that citing legal authority is a weak proxy for genuine reasoning and that verdicts remain vulnerable to manipulation.
Legal Research Bench (LRB) is a new benchmark comprising 413 open-ended U.S. legal research questions, each paired with a gold answer, supporting authorities, and a binary grading rubric. The study evaluates thirteen advanced language‑model agents using web search, case‑law search, page parsing, and retrieval tools, scoring responses only when all required criteria are met and cited authorities verify. Results show that even the best model, Claude Opus 4.8, achieves full correctness on only 42.9% of questions, with performance varying by legal area and task complexity, and no clear link between more tool calls or inference cost and higher accuracy.
arXiv:2610.15634v1 Announce Type: new Abstract: Sycophancy in large language models risks amplifying unsupported claims or beliefs across the knowledge ecosystem as these systems increasingly contrib...
The paper examines how automated safety systems unevenly flag harmful content, disproportionately labeling African American English as offensive and misclassifying content from drag queens and white nationalists. It identifies seven failure modes rooted in vocabulary-based harm detection and proposes an evaluation protocol that highlights gaps in performance across languages. Applied to five open systems in eleven languages, the study finds that none achieve native-speaker accuracy, and higher capability often increases discrimination, widening the disparity between best- and worst-served language communities.
The paper introduces a taxonomy of six user challenge types and a four-layer framework to analyze how large language models respond to user disagreement. Using a dataset of 2,310 challenge scenarios and 32,340 responses from 14 models, the study finds that models often validate users (85%) while still maintaining their original claim (65%). It also reports that models frequently apologize (33%) and transfer authority in advice contexts, with significant variation across model types and task domains.
arXiv:2607. 01153v1 Announce Type: cross Abstract: Safety evaluations for language models increasingly depend on judgments about ambiguous natural-language behaviour: whether a model has followed an instruction, refused appropriately, complied with a policy, resisted an embedded command, or misreported progress in an agentic task.
The paper introduces ComplianceAgent, a legal compliance sub‑agent within the COMPASS multi‑agent framework, which uses a large language model to evaluate texts on a fixed rubric across digital sovereignty, carbon impact, compliance, and ethics. The agent grounds its judgments in a reference legal corpus, and if a text scores at or below a critical threshold, it initiates a corrective chain that includes self‑healing rewrites, a compliance label, an explainability report, and escalation to a human reviewer. The authors evaluate the agent in a Canadian and Quebec context on a corpus of thirteen legal instruments and a hand‑annotated gold standard of fifteen texts, demonstrating perfect separation of conforming versus non‑conforming texts, stable scores under repeated queries, and relevant passage retrieval.