In re Marla C. Martin (Bankr. N.D. Ill.): $5,500 Rule 9011 Sanction for a ChatGPT-Fabricated Brief

Verification status: PENDING — checker-confirmed at confidence 1.000, awaiting HUMAN client-confirmation. Not verified. The checker independently re-verified the byte-tie, all quotes, and the two-newsroom independence (Reason/Volokh + Bloomberg Law) and graduated the story to pending; a green badge still requires a human confirming the docket posture (payment/appeal/NCBJ-course/IARDC).

This is a court-adjudicated honest-negative: a cautionary AI-governance story sourced from a signed federal court order, not a vendor deployment win.

The problem

Marla C. Martin, a serial Chapter 13 debtor (this was her eighth bankruptcy case), was represented by Thomas E. Nield of The Semrad Law Firm, LLC. Opposing creditor Corona Investments, LLC’s objection to the feasibility of her plan, Nield filed a reply brief (Dkt. No. 51) that cited four bankruptcy decisions — In re Montoya, In re Jager, In re Coleman, and In re Russell — with quotations that did not appear in the real opinions. One of the four, In re Jager, did not exist at all; the other three were real cases misattributed by district or year and cited for propositions they did not support. On June 11, 2025, Judge Michael B. Slade issued an order directing Nield and Semrad “to show cause why they should not be sanctioned for filing a brief containing fake quotations and nonexistent authority manufactured by artificial intelligence.”

What was built

Nothing was built — this is a misuse case. Asked at the June 10, 2025 hearing whether he had used AI, Nield admitted: “I ran it through AI to some extent, but I didn’t think that the citation was wrong.” Semrad’s response to the show-cause order identified the tool: Nield’s “use of ChatGPT for this purpose was ‘outside of the firm’s research protocol.’” The firm said it had since created a formal AI policy, required CLE training in the ethical use of AI, and withdrawn its fee application in the case. The court held that “any lawyer unaware that using generative AI platforms to do legal research is playing with fire is living in a cloud.”

The outcome

In a Memorandum Opinion issued July 18, 2025 (Case No. 24 B 13368), Judge Slade found that Nield “violated Federal Rule 9011” and that a “modest, joint-and-several sanction of $5,500, paid to the Clerk of the Bankruptcy Court, along with a requirement that Mr. Nield and another senior Semrad attorney attend an upcoming course on the dangers of AI scheduled for the National Conference of Bankruptcy Judges (NCBJ) annual meeting in September, is the least harsh sanction that will appropriately address counsel’s conduct.” The U.S. Trustee had argued for up to $15,000; the court imposed the lesser sum. The $5,500 is a single joint-and-several figure on Nield and the firm — not a per-attorney amount — and the NCBJ course, the IARDC self-report, and the withdrawn § 329 fee application are additional, non-monetary elements, not separate fines. The independent origin of every figure is a signed federal court order, the same court-adjudicated class as Mata v. Avianca, Wadsworth v. Walmart and Gauthier v. Goodyear (all cited in this opinion).

Path to green

The $5,500 sanction, the Rule 9011 violation, the ChatGPT attribution, and the NCBJ-course requirement are already established by the signed, published opinion — no client-confirmation call is needed to make them true. What remains open (and is asserted by no claim) is docket-only: whether the $5,500 was paid, whether the two attorneys attended the NCBJ course, the IARDC disposition of Nield’s self-report, and whether the sanction was appealed. A human confirming that docket posture is the step that would earn a green badge, if TIN carries honest-negatives. No agent may set verified.