Legal QA

QA: Question Answering

Overview

Legal question answering studies how AI systems can answer questions about laws, cases, and regulations while accounting for the facts and jurisdiction involved. Common approaches retrieve authoritative passages with keyword, dense, or hybrid search, then generate answers; newer systems add legal-structure-aware indexing, multi-agent reasoning, and claim-by-claim citation checks. Benchmarks increasingly test realistic consultations, citation support, missing facts, and changes in the law over time, rather than only multiple-choice accuracy. A central challenge is that retrieval can improve grounding but does not guarantee correct answers: systems may misuse supplied law, cite unsupported claims, or fail to ask for clarification or abstain when evidence is insufficient.

Latest papers

CardsList

64 papers8 papers in the last four weeks, down 11% on the four weeks before. 0.1% of all new papers.

  1. Benchmarking Vietnamese Legal Knowledge of Large Language Models

    Dec 16, 2025Nguyen Tien Dong, Minh-Anh Nguyen, Thanh Dat Hoang +6Language Model Reasoning EvaluationLegal Reasoning

  2. L-MARS: Legal Multi-Agent System with Agentic Search and Citation-Faithfulness Audit

    Aug 31, 2025Boqin Yuan, Ziqi WangLegal Citation VerificationAgentic Search