In a decision of first impression, the U.S. Bankruptcy Court for the Northern District of Illinois (the court) imposed sanctions on a debtor’s counsel and his law firm for filing a brief that included fabricated citations to case law and nonexistent quotations that were generated by artificial intelligence (AI). In issuing its opinion, the court provides practitioners a clear benchmark on the uses and limits of leveraging artificial intelligence in legal filings.

Relevant Background

In September 2024, the debtor, Marla Martin (the debtor), arrived before the court for what was her eighth Chapter 13 bankruptcy case (the 2024 case). In the 2024 case, the debtor’s primary liabilities were unpaid real estate taxes owed on her home, leading to a tax lien acquired by Corona Investments, LLC (Corona). In an attempt to deal with Corona’s secured claim, the debtor prepared multiple Chapter 13 plans that proposed various levels of monthly payments to Corona. Unfortunately for the debtor, the key defect with these plans, as pointed out in objections lodged by Corona (and by the court itself), was that the plans proposed to pay Corona monthly payments that exceeded the debtor’s alleged monthly disposable income, at least according to the debtor’s schedules (which were filed under oath). In response to Corona’s objections, the debtor and its counsel argued, among other things, that Corona lacked standing to object.

The court extensively reviewed the debtor’s counsel’s brief and found that it was rife with citations and quotations of legal authority that were either inaccurate or that did not exist at all. The court went so far as to include a table in its decision that compared the citations and quotations at issue to what actually exists in the alleged supporting case law. In total, the court found that debtor’s counsel cited at least four cases for a proposition of law that did not exist for what the debtor’s counsel alleged, and worse than that, allegedly quoted statements in the debtor’s brief were not in the underlying decisions nor in any other decision the court could locate.

Following its review, the court raised its concerns at a hearing attended by the debtor’s counsel and asked the counsel directly whether it used any form of artificial intelligence (AI) to assist in preparing the brief. The debtor’s counsel admitted to using AI for legal research and that he did not verify the results as he assumed that the AI would not fabricate quotes. As a result, the court issued a show cause order as to why the debtor’s counsel and his law firm should not be sanctioned for filing a brief that included incorrect or nonexistent citations to case law that were hallucinated by AI.

Rule 9011 Violation and Standard

At the order to show cause hearing, the court analyzed Federal Rule of Bankruptcy Procedure 9011, which mandates that legal contentions must be based on existing law or a nonfrivolous argument after reasonable inquiry. Although, as the court noted, what happened in this case does not appear to have been previously addressed by any other bankruptcy court in a published decision, there is a growing body of case law from district courts addressing similar situations. The court reviewed those decisions and concluded that the holdings of those decisions were: “both uniform and highly persuasive;” uniform in requiring that the duties imposed by Federal Rule of Civil Procedure 11 (and by extension, Bankruptcy Rule 9011) included the requirement that attorneys verify the existence and validity of any cited legal authorities on which they rely; and uniform in requiring that these duties could not be outsourced to AI.

As a result, the court noted that sanctions for this type of conduct are appropriate. According to the court, available sanctions include nonmonetary directives, penalties payable to the court, and fee shifting. In addition, the court also found that law firms are to be held jointly responsible for their lawyers conduct, absent exceptional circumstances. However, sanctions generally must be tailored to the offense and “limited to what suffices to deter.”

Sanctions Imposed and Rationale

The debtor’s counsel and its law firm asked the court to forego imposing sanctions in light of what they described to be concessions already granted, namely that they agreed to forego compensation in this case and that they readily admitted to the misconduct in this case. While the court acknowledged their candor, it asserted that “there must be consequences” as citations and reliance on fake, AI-generated authority, even if unintentional, “shatters credibility” and “imposes many harms.” In fact, the court emphasized that at this point in time, in 2025, the risk of AI hallucinations is (and should be) widely known in the profession, referencing various widely available court policies and decisions. To that end, the debtor’s counsel’s allegations that he was unaware that AI could result in fake citations was particularly “troubling” to the court. Indeed, the court reiterated that lawyers hold an ethical obligation to not only review the cases they cite and present to a court (regardless of where those cases are pulled from), but to further understand and keep abreast of developments in technology germane to their practice.

Using all of this guidance as a roadmap, this decision sets a clear, practitioner-focused benchmark: unverified AI-generated authority in court filings violates Rule 9011 and violators will likely be sanctioned.

Reprinted with permission from The Legal Intelligencer, © ALM Media Properties LLC. All rights reserved.