Courts on AI / Self-represented litigants using AI
Webb v. Sullivan County Assessor
Indiana Tax Court · Other · IN · · No. 25T-TA-00029 · opinion
The Indiana Tax Court suspects self-represented taxpayers used generative AI in a motion citing real rules for incorrect propositions. The court recognizes benefits for self-represented parties but warns that all litigants must verify sources and holdings.
- Disposition
- The court denied the pro se taxpayers’ motion to supplement the record and warned about AI citation accuracy.
What the court wrote
Verbatim from the court's document (public domain). Ellipses mark omitted text; nothing is paraphrased.
The Webbs refer to various court rules in their motion to support their request to supplement the record. While the rules the Webbs cite do exist, they plainly do not stand for any of the propositions that the Webbs advance. This leads the Court to believe that the Webbs used generative AI in producing their Motion to Supplement the Record. The Court recognizes the benefits of generative AI, especially for self- represented parties like the Webbs, and does not wish to dissuade its responsible use. But that does not relieve attorneys and self-represented litigants of their duty to independently verify the authenticity and accuracy of authoritative sources cited to the Court. Courts “must be able to rely on the authenticity of the authorities cited by the parties to make just decisions.” Williams v. Kirch, 268 N.E.3d 284, 288 (Ind. Ct. App. 2025). Generative AI tools are not reliable for legal citation and “frequently generate fictitious case names and citations.” Wilcox v. Gingrinch, 274 N.E.3d 1269, 1277 (Ind. Ct. App. 2026). Even when such tools do cite authentic cases, they frequently mischaracterize their meaning or essential holdings. See, e.g., Sanders v. United States, 176 Fed. Cl. 163, 170 (Fed. Cl. 2025) (“AI programs can generate fake case citations and other legal misstatements.”). See generally Isabel Gottlieb & Isaiah Poritz, Popular AI Chatbots Found to Give Error-Ridden Legal Answers, BLOOMBERG L., (Jan. 12, 2024), https://news.bloomberglaw.com/us-law-week/legal-errors-by-top-ai-models- alarmingly-prevalent-study-says (discussing research from Stanford University that found, out of 200,000 questions, “[l]arge language models hallucinate at least 75% of the time when answering questions about a court’s core ruling”). While this technology will undoubtedly be refined and improved in the course of time, litigants must continue to exercise care to ensure that the information gleaned from such sources is reliable and accurate.
Authorities cited in the passage: Williams v. Kirch, 268 N.E.3d 284 (Ind. Ct. App. 2025) · Wilcox v. Gingrinch, 274 N.E.3d 1269 (Ind. Ct. App. 2026) · Sanders v. United States, 176 Fed. Cl. 163 (Fed. Cl. 2025)