TriBench-Ko: Evaluating LLM Risks in Judicial Workflows

TL;DR

TriBench-Ko is a Korean benchmark designed to evaluate the risks and performance of large language models in real judicial tasks, highlighting significant challenges and areas needing caution.

Contribution

It introduces a comprehensive benchmark for assessing LLM risks in judicial workflows, covering four core legal tasks with detailed risk categories.

Findings

Many LLMs struggle with precedent retrieval.

Models often manifest significant risks like hallucination and bias.

Outputs frequently fail to capture critical legal information.

Abstract

Large language models (LLMs) are increasingly integrated into legal workflows. However, existing benchmarks primarily address proxy tasks, such as bar examination performance or classification, which fail to capture the performance and risks inherent in day-to-day judicial processes. To address this, we publicly release TriBench-Ko, a Korean benchmark designed to evaluate potential deployment risks of LLMs within the context of verified judicial task requirements. It covers four core tasks: jurisprudence summarization, precedent retrieval, legal issue extraction, and evidence analysis. It jointly assesses model behavior across multiple deployment risk categories, including inaccuracy (hallucination, omission, statutory misapplication), biases (demographic, overcompliance), inconsistencies (prompt sensitivity, non-determinism), and adjudicative overreach. Each item is structured to systematically assess both task performance and a specific risk type based on real judicial decisions. Our evaluation of a range of contemporary LLMs reveals that many models frequently manifest significant risks, most notably struggling with precedent retrieval and failing to capture critical legal information. We provide a comprehensive diagnosis of these LLMs and pinpoint critical areas where LLM-generated outputs in judicial contexts necessitate rigorous inspection and caution. Our dataset and code are available at https://github.com/holi-lab/TriBench-Ko