Fivehouse v. U.S. Department of Defense
U.S. District Court, Eastern District of North Carolina, Northern Division · E.D.N.C. · North Carolina bar guidance
Conduct
An Assistant U.S. Attorney defending a civil suit used generative AI to draft a brief with fabricated quotations and admitted it only under the court's questioning.
Consequence
Public reprimand, opinion published; no fine, given the loss of his job and a pending DOJ Office of Professional Responsibility investigation.
Lesson
Delayed disclosure of AI use turns a verification failure into a candor finding; government attorneys face the same standard.
Verified September 28, 2026
- Citation
- Fivehouse v. U.S. Dep't of Def., No. 2:25-CV-00041-M-RN, Amended Order (E.D.N.C. May 7, 2026) (Numbers, M.J.), ECF No. 131
- Decided
- May 7, 2026
Summary
In an Administrative Procedure Act suit against the Department of Defense and other federal defendants, then-Assistant United States Attorney Rudy E. Renfer filed a December 23, 2025 response brief containing fabricated quotations and misstated holdings of real decisions, including Ohio Valley Environmental Coalition v. Aracoma Coal Co. and Dow AgroSciences, LLC v. National Marine Fisheries Service, and a pin cite to a page past the end of Sierra Club v. U.S. Department of the Interior. His written submission in response to the show-cause order and his opening statement at the March 10, 2026 hearing did not mention AI. Only when the court questioned him about each fabrication did Renfer admit that when he discovered on December 22 that he had saved over his draft, he "panicked" and "used artificial intelligence to catch . . . back up," intentionally using a generative AI tool to draft the brief without verifying it. The court noted that his filing deadline was December 30, more than a week after he says he discovered the problem, and found his account not credible. The United States Attorney told the court this was the first he had heard of Renfer using AI for the brief.
- AI tool:
- Unnamed generative AI tool (admitted by the attorney only under the court's questioning at the show-cause hearing)
- Sanction amount:
- Non-monetary: public reprimand of former Assistant United States Attorney Rudy E. Renfer; opinion submitted for publication; no financial penalty
What sanction did the court impose?
In an amended order, Magistrate Judge Robert T. Numbers, II found by clear and convincing evidence that Renfer "knowingly and intentionally submitted a document to the Court that contained fabricated quotations and misstatements of case holdings," found his account of events not credible, and, in the alternative, found bad faith. Because the legal contentions in the brief were valid, the court concluded that its inherent authority was better suited than Rule 11, and it publicly reprimanded Renfer and directed the Clerk to submit the opinion for publication and the United States Attorney's Office to mail him a copy. The court declined to impose a further financial penalty, noting that Renfer's conduct had cost him his position with the United States Attorney's Office and that he is the subject of an investigation by the Department of Justice's Office of Professional Responsibility. The amended order corrects a typographical error in the original and makes no substantive changes.
Why does Fivehouse v. U.S. Department of Defense matter for law firms using AI?
Fivehouse frames its question at the outset: “what should a court do when an attorney submits a brief containing fabricated legal authority and attributes the errors to the use of generative artificial intelligence? The answer is straightforward. The court should hold the attorney accountable because he is responsible for the authorities he presents to the court, regardless of how they came to appear in the brief.”
The finding that drives the order is about candor, not technology. Renfer’s written response to the show-cause order and his opening remarks at the hearing said nothing about AI. The court drew the admission out only by questioning him on each fabrication in turn, and then found his account of how the brief came to be filed not credible. That sequence supported a finding, by clear and convincing evidence, that he acted knowingly. Because Rule 11(b) targets legal contentions and “the legal contentions in his filings were valid,” the court found its inherent authority better suited to the conduct, and added that Renfer’s service as an Assistant United States Attorney made the misconduct “particularly odious” given “a position of immense trust and power.”
The court limited itself to a public reprimand because Renfer had already lost his position and faced an Office of Professional Responsibility investigation. It signaled that the calculus is changing: dealing with the problem “will, necessarily and unfortunately, involve moving beyond admonitions and reprimands into more punitive sanctions.” Firms documenting compliance may wish to consider that disclosure timing is now itself a sanctions factor.
Implications for your firm
Operational steps a firm reading this case may wish to consider documenting. Strategic and rule-application calls belong to your firm's attorneys.
- Document a recovery procedure for lost or corrupted drafts. The attorney said he turned to generative AI after discovering he had saved over his draft, although, as the court noted, more than a week remained before the deadline.
- Train attorneys to disclose AI use at the first opportunity. The court found that Renfer's silence in his written submission and opening statement, followed by admission under questioning, supported a finding of knowing misconduct.
- Review office-wide guidance after an incident. Following the show-cause order, the U.S. Attorney issued a memo cautioning against relying on AI to draft documents, instructing staff to personally verify each quote or proposition in the actual source, and requiring two professional-responsibility courses for every attorney in the office.