Using AI in Federal Taxation Practice: Guidelines for Professionals and Clients
By Sandra D. Mertens, Esq.[1]
Generative AI programs such as Chat GPT have been around for several years now and legal clients often conduct their own research using generative AI or attempt to shortcut legal fees by asking AI to draft a court pleading or contract. Yet misuse of AI is in the news regularly, with numerous warnings and sanctions imposed by the Courts as they adapt to the use (and misuse) of AI by attorneys and pro se litigants alike. The IRS and Tax Court are likewise adapting and developing policies and guidelines for practitioners and taxpayers.
Recent Tax Court Decisions on AI
The Tax Court was faced with AI use and hallucinations in the case Gary Thomas v. Commissioner, Docket No. 10795-22. As explained in its October 23, 2024 Order, the Tax Court “noticed that some of the authorities cited in petitioner’s Pretrial Memorandum did not exist, evidencing possible AI hallucinations.” The Court held a hearing to provide petitioner’s counsel an opportunity to clarify, but petitioner’s counsel claimed that someone else prepared the document and she did not review it.
The Court noted that the Pretrial Memorandum violated Federal Rule of Civil Procedure 11 because the attorney signed it without reading it and was unable to certify that it was well-grounded in fact and warranted by existing law. Therefore, the Court struck the Pretrial Memorandum and issued a 5-page opinion as an instruction to tax professionals and pro se parties in the use of AI.
The Court explained the problem: “none of the cases referenced in [select] passages quoted above exist as cited. Moreover, neither the named cases nor their accompanying citations stand for the propositions for which they were cited.” Some of the case names existed under different citations, while other citations referred to other cases; but overall, the cases were not relevant or helpful, nor was the petitioner’s use of them accurate.
While AI has the potential to increase access to justice and provide resources to meet the needs of the court system, it can result in “hallucinations” which may include citations to non-existent cases. When this happens, the Court’s resources are diverted in attempts to track down the erroneous citation. As such, the Court decided that the circumstances of the case warranted “a minimal sanction,” and thus took “the symbolic action of deeming the Pretrial Memorandum to be stricken.”
By February 2026, the Tax Court had enough. In Peter L. Clinco, Deceased v. Commissioner, T.C. Memo. 2026-16 (Feb. 9, 2026), the tax issue was simple: the IRS believed the taxpayers had underreported gross receipts on Schedule C from their family restaurant and contested the taxpayers’ depreciation deduction for certain rental properties. After an audit, the IRS issued a notice of deficiency which was signed electronically and not manually (i.e., in ink) by an IRS employee. The taxpayers challenged the deficiency notice on this basis, but the Tax Court, citing both case law and the Internal Revenue Manual, quickly sided with the IRS that even an unsigned Notice of Deficiency is valid, an issue which “has been settled since before World War II.”
The Tax Court then turned to the “Fabricated Case Citations” submitted by the taxpayers and their attorney. Of the four cited cases, three appeared to be hallucinations generated by a large language model AI. Two of the cases did not exist as cited, while the third case existed under a different citation but did not pertain to the relevant issues. Although the IRS had raised the questionable citations in its answering brief, the petitioner “chose not to clarify their origins in his reply.” Noting the “smorgasbord of cases condemning fake citations,” the Tax Court concluded: “Their presence is unacceptable.”
“Submitting a brief with fictitious caselaw is a recipe for sanctions and a clear violation of Rule 11(b) of the Federal Rules of Civil Procedure.” The Tax Court censured the attorney and issued a warning: “A bit of embarrassment for failure to citecheck [and] failure to ‘fess up . . . . But courts have begun to more seriously sanction [parties] who use AI as a shortcut in drafting. . . . Tax Court has not done so. Yet.”
Courts Are Imposing Monetary Sanctions for AI Misuse
Courts are beginning to impose monetary and non-monetary sanctions on parties and pro-se litigants who misuse AI and cite non-existent cases. In two recent unpublished decisions, Ifeoma Delliane Chinedu Obi v. Cook County, Illinois, et al., Case No. 1:25-CV-03096, Dkt. 97 (N.D.Ill. Apr. 9, 2026) and 2026 WL 1678232 (N.D.Ill. June 10, 2026) (slip copy), the Chicago federal court imposed monetary sanctions on the pro se petitioner who filed a frivolous motion. While ordinarily some leeway might be afforded to a pro se litigant, here “Plaintiff’s egregious, repeated, and ongoing Rule 11 violations,” the Court found, “foreclose any such possibility.”
The Court continued: “Plaintiff generated each brief using AI. Plaintiff’s motion is riddled with AI hallucinations, made up cases, quotes, and statements of law and fact.” Further, “Plaintiff’s motion is so replete with fabricated law, quotes, and statements that there is virtually no cogent argument actually supported by real law. . . . This is not the first time Plaintiff has done this. In a prior filing, Plaintiff’s brief contained at least 17 instances of fake cases, quotes, and statements of law and fact from AI hallucinations.” In that instance, the Court “gave Plaintiff grace—Plaintiff has exhausted that leniency.” Citing Rule 11, the Court imposed a $5,000 monetary sanction and struck her motion.
New IRS Guidelines on AI
On June 24, 2026, the IRS Office of Professional Responsibility introduced guidelines for responsible AI use in federal tax practice. Recognizing that “[v]irtually all professional tax firms use some form of AI, whether they are aware of it or not,” and referencing products such as Thomason Reuters’ Westlaw Edge, Bloomberg Tax, and Lexis-Nexis, the IRS noted AI’s “transformative potential includes cost savings, rapid data analysis and, for government tax administrators, in particular the IRS, advanced applications such as fraud detection and audit risk assessment.”
The IRS also identified a number of concerns, including fabricated outputs (i.e., hallucinations), bias, lack of transparency, privacy, confidentiality, and data protection. The guidelines outlined Circular 230 provisions relevant to AI use by tax practitioners, including the requirement of “due diligence” in preparing, approving, and filing tax returns, documents, affidavits, and other papers relating to IRS matters. Due diligence requires verifying the accuracy of facts, citations, and calculations produced by AI. Taxpayers and their hired professionals cannot rely solely on AI; human scrutiny and editing are essential to ensure correctness and compliance with IRS expectations.
Positions and arguments should be based on reasonable factual and legal assumptions, which the IRS concludes means that taxpayers may not rely on AI projections or representations without verification. Citations must be checked and cases read. Financial forecasts and formulas must be confirmed. Therefore, taxpayers and their representatives must independently authenticate all factual and legal information rather than blindly rely on AI outputs.
AI Best Practices for Tax Practitioners
The IRS guidelines provide the following best practices for tax practitioners:
- Identify, understand, and stay updated on any relevant federal or state-specific laws, regulations, and guidance that pertain to your professional activities.
- Establish secure AI data handling protocols and access controls.
- Document AI usage and verification processes.
- Foster transparency and accountability in all AI practices.
- Prepare clear procedures for handling breaches or errors.
- Provide necessary staff training.
- Vet third parties’ AI offerings at the time of or before purchasing.
- Never upload sensitive data to unsecured sites.
- Treat the written text that AI generates as drafts.
- Review the resulting documents thoroughly for factual and legal accuracy
(e.g., always check citations) and any problematic bias.
Even the IRS recognizes the powerful potential in AI to streamline routine tasks, enhance research, and provide valuable insights, but cautions that final decisions must always rest with a qualified tax professional who understands the complexities of tax law and ethical standards.
SFBBG’s Tax Services Department Remains at the Forefront of the AI Field
While many legal and accounting firms struggle to adapt to changes in technology, SFBBG has incorporated AI into its tax and legal services to harness the benefits of providing cost savings to clients while also ensuring any AI results are thoroughly verified and appropriately utilized. Clients should expect their attorneys and tax professionals to consider how AI can streamline or shortcut repeated processes or lengthy analyses. Professionals who refuse to adapt do a disservice to their clients.
Importantly, SFBBG’s attorneys verify all AI results to ensure Court filings do not contain hallucinations or other improper content. While clients may submit AI drafts in an attempt to reduce fees, most AI platforms cannot replace an attorney’s vast experience and knowledge required for superior advocacy. Often, AI drafts are not usable at all because they do not conform to a required format, violate applicable rules, contain poorly-developed or inapplicable arguments or content., Nothing can replace an experienced attorney’s careful analysis and creative arguments, increasingly built with skillful AI usage.
SFBBG provides federal, state, and local tax services to individuals, businesses, estates, and trusts dealing with IRS notices or complex tax-related issues. Our attorneys represent clients before the IRS, Illinois Department of Revenue, U.S. Tax Court, and Illinois Tax Tribunal, among others. Understanding when and how AI should be used in these scenarios is vital to our client representation.
[1] Sandra D. Mertens is a Partner at Schoenberg Finkel Beederman Bell Glazer LLC, and may be contacted at (312) 648-2300 or Sandra.Mertens@SFBBG.com.
