📄 中文摘要
在税务简报中引用不存在的案例是一个结构性问题，而非风格性问题。优化为听起来有说服力的语言模型（LLMs）可能会生成看似有效但实际上没有任何报告或案号依据的“Clinco诉专员”类权威引用。在税务等以先例为驱动、面临制裁风险的领域，这种失误是非常严重的。税务诉讼律师、内部税务团队和上诉专家必须了解语言模型如何产生幻觉、如何检测虚构引用以及如何管理AI的使用，这已成为核心能力。
Clinco诉专员：税务法庭中的AI幻觉与虚构法律引用
📄 English Summary
Clinco V Commissioner Tax Court Ai Hallucinations And Fictitious Legal Citations
Citing non-existent cases in tax briefs represents a structural issue rather than a stylistic one. Language models (LLMs) optimized for persuasive language can generate authorities like 'Clinco v. Commissioner' that appear valid but lack any basis in actual reports or dockets. This serious failure is particularly concerning in precedent-driven domains like tax, where the risk of sanctions is present. For tax litigators, in-house tax teams, and appellate specialists, understanding how LLMs hallucinate, detecting fabricated citations, and governing AI usage has become a core competency.