Late last week, the Seventh Circuit held that a private property tax purchaser holds a secured claim that qualifies as a tax claim under § 511(a). It thus also held that if this secured claim is to be paid during a chapter 13 plan, § 511(a) provides that “applicable nonbankruptcy law” supplies the interest rate. Bernardo Romero v. Corona Investments, LLC, 25-02021 (7th Cir. July 16, 2026). This holding is not what is most remarkable about this case, even though there was a dissent contesting the characterization of a private purchaser as holding a claim under § 511(a) and thus the application of an interest rate other than Till.
The more remarkable aspect of the case and opinion is that the attorney for the tax purchaser apparently used AI to draft its brief, which resulted in the inclusion of AI hallucinated quotes. The debtor’s attorney brought this to the court’s attention with a motion to strike portions of the brief. The attorney, being candid, noted that the AI hallucinations did not materially affect the presentation of the appeal. So the majority opinion ends by merely “lodg[ing] a general reminder that the court expects members of our bar to exercise care and diligence in preparing their briefs to ensure complete factual and legal accuracy.” The dissent possibly wanted to go further. It highlights that the tax purchaser’s brief “included an astonishing number of erroneous and even hallucinated citations,” but ultimately merely notes that the hallucinations “made this court’s work more difficult than it should have been. We should expect and insist on better, more professional performance.”
Sure, sanctions have been imposed in other recent cases with hallucinated citations (including in the Seventh Circuit) and could have been appropriate here. Instead, what this law firm received was the lightest tap on the wrist by being named, both the individual attorney and the firm, in the dissent only. One firm uses AI seemingly without checking the results (which I’m confident in writing is a professionally irresponsibly use of AI) and the other not only catches it, but also professionally appropriately brings it to the court’s attention, including sketching out the scope of the problems introduced by opposing counsel’s irresponsible use of AI. I’m gearing up to teach contracts this fall. And I’m incorporating more discussion of the use (and misuse) of GenAI tools than I ever have before. What does this teach students (and attorneys) about the ramifications of over-dependence on AI?
