Can LLMs Identify Tax Abuse?

Andrew Blair-Stanek, Nils Holzenberger, Benjamin Van Durme
8/10/2025
q-fin.CPcs.AI

No official code found. Best implementation: BlairStanek/LLM-Tax-Abuse (5 stars, last commit 2025-10-20).

Abstract

We investigate whether large language models can discover and analyze U.S. tax-minimization strategies. This real-world domain challenges even seasoned human experts, and progress can reduce tax revenue lost from well-advised, wealthy taxpayers. We evaluate the most advanced LLMs on their ability to (1) interpret and verify tax strategies, (2) fill in gaps in partially specified strategies, and (3) generate complete, end-to-end strategies from scratch. This domain should be of particular interest to the LLM reasoning community: unlike synthetic challenge problems or scientific reasoning tasks, U.S. tax law involves navigating hundreds of thousands of pages of statutes, case law, and administrative guidance, all updated regularly. Notably, LLM-based reasoning identified an entirely novel tax strategy, highlighting these models' potential to revolutionize tax agencies' fight against tax abuse.

View on arXivView PDF

Code Implementations(1)

BlairStanek/LLM-Tax-AbuseCommunity53%

Code and Tax-Strategy Data from Paper 'Can LLMs Identify Tax Abuse?'

51PythonAug 5, 20251 years ago

Similar Papers

Taxation Perspectives from Large Language Models: A Case Study on Additional Tax Penalties

88% match2025

Using Large Language Models for Legal Decision-Making in Austrian Value-Added Tax Law: An Experimental Study

85% match2025

Large Language Models as Tax Attorneys: A Case Study in Legal Capabilities Emergence

84% match2023

ASDA: Automated Skill Distillation and Adaptation for Financial Reasoning

83% match1 repo72026

An LLM Agentic Approach for Legal-Critical Software: A Case Study for Tax Prep Software

82% match2025

Language Models and Logic Programs for Trustworthy Tax Reasoning

82% match2025

Cite this paper

@article{blairstanek2025llms,
  title  = {Can LLMs Identify Tax Abuse?},
  author = {Andrew Blair-Stanek and Nils Holzenberger and Benjamin Van Durme},
  year   = {2025},
  eprint = {2508.20097},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2508.20097v2}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2508.20097.svg)](https://codeofpaper.com/paper/2508.20097)

Discussion