Fivehouse v. U.S. Dept. of Defense (E.D.N.C.): a federal prosecutor is publicly reprimanded for a brief with AI-fabricated quotations

This is a court-adjudicated honest-negative: a cautionary AI-governance case sourced from a signed federal order, not a vendor deployment win. What distinguishes it is who was sanctioned: a federal prosecutor, not private counsel source.

The problem

Derence V. Fivehouse, a retired Air Force colonel representing himself, sued the U.S. Department of Defense under the Administrative Procedure Act, challenging a change to medication covered under the TRICARE program source. Opposing Fivehouse’s motion to supplement the administrative record, the Government’s lawyer, then-Assistant U.S. Attorney Rudy E. Renfer, filed a response brief. Fivehouse read it, checked the authorities, and told the court the brief included fabricated quotations and misrepresented the holdings of cases source. The court reviewed and agreed, issuing a show-cause order in March 2026 source.

What was built

Nothing was built; this is a misuse case. Renfer testified that after he inadvertently saved another filing over his draft, he discovered the loss on December 22, 2025, panicked, and used artificial intelligence to catch back up, then did not verify the quotations source. In the court’s words: “These errors arose because Renfer used a generative AI tool to draft the brief but failed to verify the accuracy of its contents” source. At the hearing in Raleigh he said the decision to use AI was the worst of his career source. The order names no specific AI product. Do not attribute it to any vendor.

The outcome

In an 18-page Order filed April 28, 2026 (No. 2:25-CV-00041-M, Document 129), U.S. Magistrate Judge Robert T. Numbers II sanctioned Renfer under the court’s inherent authority source. The court found by clear and convincing evidence that Renfer knowingly and intentionally submitted a document containing fabricated quotations and misstatements of case holdings, rejecting his position that the errors were inadvertent source.

The sanction is a public reprimand and no monetary fine. The court explained why: Renfer’s conduct had already cost him his position, and that loss of employment imposed a financial burden well beyond the fines courts typically impose in AI-related matters, citing for comparison Wadsworth v. Walmart, Gardner v. Combs, and Mata v. Avianca source. The court also ordered its opinion submitted for publication, and the case now appears in practitioner surveys of 2026 generative-AI sanctions source.

The conflict in the record

Sources do not agree on how Renfer left the U.S. Attorney’s Office, and this case file does not merge them. The order records that Renfer told the court he had decided to separate from the office source. Bloomberg Law reports the Department of Justice fired him the day after the hearing source. Law360’s account frames the departure as a resignation source. What is not in dispute across all three: he no longer holds the post, and the court weighed that fact when declining a further penalty. The employment action is an Executive-Branch matter, not a court-imposed sanction.

How this was verified

Method: Independent audit of the public record. The primary document, the signed 18-page Order (Document 129, E.D.N.C. No. 2:25-CV-00041-M, filed 2026-04-28), was read directly and is self-authenticating Tier 1 evidence; the docket was confirmed on CourtListener. Both load-bearing facts (the AI attribution and the public-reprimand sanction) are independently carried by Bloomberg Law, ABA Journal and Law360, so no claim here rests on the court document alone. The one point on which sources diverge, how the attorney left the U.S. Attorney’s Office, is shown as a conflict above rather than resolved. No subject of this story was contacted; TIN does not seek confirmation from the parties it audits.

Verification date: 2026-08-14

Related case files

• Wadsworth v. Walmart: cited by this court as a fine-level comparator.
• Gardner v. Combs: the other comparator named in the order.
• U.S. v. Hayes: the closest parallel involving government counsel.

Sources

1. Tier 1 (primary court document): U.S. District Court, E.D.N.C., Fivehouse v. U.S. Department of Defense, No. 2:25-CV-00041-M, Order (Doc. 129), 2026-04-28. https://www.courthousenews.com/wp-content/uploads/2026/04/fivehouse-department-of-defense-renfer-reprimand.pdf
2. Tier 1 (primary docket): CourtListener, Fivehouse v. U.S. Department of Defense, 2:25-cv-00041 docket. https://www.courtlistener.com/docket/71231282/fivehouse-v-us-department-of-defense/
3. Tier 2 (independent trade press): Bloomberg Law, “Court Rebukes Fired DOJ Lawyer for ‘Odious’ and False AI Filings,” 2026. https://news.bloomberglaw.com/us-law-week/court-rebukes-fired-doj-lawyer-for-odious-and-false-ai-filings
4. Tier 2 (independent trade press): Law360, “Prosecutor Resigns, Judge Shows Slide Deck On AI Errors,” 2026. https://www.law360.com/articles/2450187/prosecutor-resigns-judge-shows-slide-deck-on-ai-errors
5. Tier 2 (independent legal press): FindLaw, “DOJ Attorney’s AI-Generated Brief Sparks Sanctions Threat After Pro Se Plaintiff Uncovers Fabricated Quotes,” 2026. https://www.findlaw.com/legalblogs/practice-of-law/doj-attorneys-ai-generated-brief-sparks-sanctions-threat-after-pro-se-plaintiff-uncovers-fabricated-quotes/
6. Tier 3 (practitioner survey / context): Norton Rose Fulbright, “AI in litigation: Update on Gen AI sanctions in 2026.” https://www.nortonrosefulbright.com/en-us/knowledge/publications/792d8bf3/ai-in-litigation-update-on-gen-ai-sanctions-in-2026
7. Tier 3 (contemporaneous local reporting): The Chetson Firm, “Federal Judge Orders AUSA to Show Cause After Fabricated Citations Found in Court Filings,” 2026-03. https://chetson.com/news/ausa-fabricated-citations-show-cause-ednc/