Xu v. Nasser (In re Nasser)
United States Bankruptcy Court, Eastern District of Michigan, Southern Division · Bankr. E.D. Mich. · Michigan bar guidance
Verified April 26, 2026
- Citation
- Xu v. Nasser (In re Nasser), Adv. Proc. No. 25-04091, Case No. 25-44069 (Bankr. E.D. Mich. Aug. 15, 2025) (Hage, J.)
- Decided
- August 15, 2025
Summary
In an adversary proceeding within debtor Marb Nasser's chapter 13 case, Judge Paul R. Hage flagged in footnote 3 of his opinion on the motion to dismiss that plaintiffs' Response in Opposition contained material errors. Plaintiffs' counsel filed a Notice of Errata acknowledging that three statutory provisions presented as block quotations were in fact paraphrased and materially different from the actual statutory text, and the court separately identified language attributed to the Complaint as direct quotation that did not appear in the Complaint at all. The court stated it was "concerned that Plaintiffs may have used generative artificial intelligence to assist in drafting the Response" and warned counsel that future errors of this nature would not be tolerated.
- AI tool:
- Unspecified generative AI (suspected; not confirmed by counsel)
What sanction did the court impose?
No monetary sanction, no referral, no formal Rule 11 finding. The court issued an on-the-record warning to plaintiffs' counsel via footnote 3 of the opinion, declining to take further action "at this time" but cautioning that future errors of this nature would not be tolerated. The motion to dismiss was granted in part (Count II under section 523(a)(4) dismissed) and denied in part (Count I under section 523(a)(2)(A) survived); Count III was withdrawn.
Why does Xu v. Nasser (In re Nasser) matter for law firms using AI?
In re Nasser is a useful counterpoint to the more dramatic AI-hallucination sanctions cases: there is no fabricated case citation, no Rule 11 referral, and no dollar figure. What there is, is a federal judge devoting a footnote to the suspicion that counsel ran statutory text and complaint quotations through a generative tool, and treating the resulting inaccuracies as a credibility problem for the entire brief. For a managing partner, the lesson is that the reputational floor is well above the sanctions floor: a judge who suspects AI-assisted sloppiness can flag it permanently in a published opinion without ever issuing a sanction, and that footnote travels with counsel into every future appearance.
Sources
Further reading
Source PDF is a Westlaw printout mirrored from the Damien Charlotin hallucination database. We are working to add the underlying court docket (PACER, CourtListener, or court website) as a second source.
- The specific generative AI tool used was not identified in the order; the court inferred AI use from the pattern of paraphrased statutory text passed off as block quotations and from quoted language attributed to the Complaint that did not appear there. Counsel did not confirm AI use on the record cited in the opinion.
- The identity of the specific attorney sanctioned (plaintiffs' counsel of record on the Response and Notice of Errata) is not extracted from this opinion; the docket would identify counsel. The opinion warns "counsel" generally without naming the lawyer.