# A Federal Court Fined a Public Defender $1,500 for a Fake Citation — and Refused to Say AI Wrote It > United States v. Hayes is the case that shows where AI blame stops. A federal magistrate judge found the fictitious citation in a defender's brief had "all the markings of a hallucinated case created by generative artificial intelligence (AI) tools such as ChatGPT and Google Bard" — then held she "need not make any finding" that AI was used, while the attorney denied ever using it. The $1,500 sanction is court-adjudicated and was paid. The AI causation is not established by anyone. - Verification status: verified - Case type: deployment - Provider: None identified — the court made no finding on AI use and the sanctioned attorney denies using AI - Client: U.S. District Court for the Eastern District of California (Hon. Chi Soo Kim, Magistrate Judge) — United States v. Daragh Finbar Hayes, No. 2:24-cr-0280-DJC, Courts / legal (AI-governance honest-negative) (named) - Sector: legal-services / US / cross - Verified on: 2026-07-25 - Canonical URL: https://theinternetninja.com/stories/us-v-hayes-federal-defender-1500-sanction-fictitious-citation-court-declines-to-find-ai-use-2025/ - Source: The Internet Ninja (theinternetninja.com), independent verified-proof platform ## Verification method Every figure is taken verbatim from signed orders of the U.S. District Court for the Eastern District of California (ECF Nos. 50, 62 and 83 in No. 2:24-cr-0280-DJC) fetched live from storage.courtlistener.com and byte-tied to Internet Archive captures by SHA-1; corroborated by the court's own docket text and by two independent secondaries (Eugene Volokh, Reason, 2025-01-23; Mealey's Artificial Intelligence / LexisNexis, 2025-01-24). ## Full case file **Verification status: PENDING — awaiting a human gate, and NOT verified.** Every figure below is Tier 1 from a signed federal court order and doubly corroborated, but one thing is still open and it is not a sourcing question: see "What we cannot say" — the sanction is under an undecided de novo appeal and we cannot reach a ruling. Do not publish before the E.D. Cal. clerk or PACER confirms whether it survived. ## The problem On 21 November 2024, Assistant Federal Defender Andrew Francisco filed a motion to unseal in a federal criminal prosecution in Sacramento. Its principal authority was `United States v. Harris, 761 F. Supp. 409, 414 (D.D.C. 1991)`, quoted for the proposition that "The legislative history indicates that Congress intended to shield the names and addresses of child victims and witnesses from public disclosure." The prosecution checked it. The citation `761 F. Supp. 409` belongs to a different case entirely — *Harris v. Murray*, a 1990 prisoner civil-rights decision from the Eastern District of Virginia — and the quoted sentence returned zero results on both Westlaw and Lexis. In the court's words: > Unfortunately, "United States v. Harris, 761 F. Supp. 409, 414 (D.D.C. 1991)" is not > a real case. The reply brief did not concede it. It said the quote actually came from *United States v. Broussard* — where the sentence also does not appear. At a hearing on 9 December 2024 the court "provided Mr. Francisco with a hard copy of *Broussard*"; when he could not find the quotation in it, he suggested it came from a second *Broussard* opinion. The court had read that one too before the hearing and "informed Mr. Francisco that *Broussard* II did not contain the quotation." Across the hearing the court gave him "five distinct opportunities to respond to the Court's direct questions" — about the quotation, about *Broussard*, and finally about whether *Harris* existed at all. He did not concede it. ## What was built Nothing was built, and that is the point of this entry. The court found the citation had the shape of a machine-generated one: > "The citation has all the markings of a hallucinated case created by generative > artificial intelligence (AI) tools such as ChatGPT and Google Bard" And then declined to decide the question: > "The Court need not make any finding as to whether Mr. Francisco actually used > generative AI to draft any portion of his motion and reply, including the fictitious > case and quotation." The attorney's own account, quoted in the order, is a denial. He states he "did not use and ha[s] never used AI (artificial intelligence) to draft any of my motions," says he drafted the passage "hastily," and adds of the fake case: "It was an inadvertent drafting error I cannot begin to explain." The court found that response "inadequate and not credible" — but its finding of non-credibility is about candour, not about AI. So the record has an AI-shaped hole in it and no one filled it. No tool, no vendor, no prompt, no admission. ## The outcome The sanction issued on 17 January 2025, under Eastern District of California Local Rule 180(e) and the court's inherent authority — **not** Rule 11, which does not reach a criminal case: > "Assistant Federal Defender Andrew Francisco, defense counsel, is personally sanctioned > in the amount of $1,500." > "the Court concludes that $1,500 in monetary sanctions issued against Mr. Francisco > personally are proportionate to his conduct" The court found the misrepresentations deliberate: > "Mr. Francisco's inaccurate and misleading statements were not inadvertent as claimed, > but knowing and made in bad faith." The consequences ran past the money. The Clerk was ordered to serve the order > "on the District of Columbia Bar, of which Mr. Francisco is a member (DC Bar No. > 1619332), and the State Bar of California" and on every district and magistrate judge in the district, because > "while federal and state courts across the country have unfortunately encountered this > issue, this is the first time it has arisen in this federal district, raising the > importance of educating and deterring the larger bar from repeating similar conduct" It was collected. The docket records, on 6 February 2025: > "RECEIPT number 200011278 for $1500.00 in Sanctions per 62 Order from Andrew Francisco." Reconsideration was denied on 9 April 2025 — "DENYING 72 & 75 Amended Motion for Reconsideration of 62 Order for Sanction." ## What we cannot say **That AI caused this.** The court refused to find it and the attorney denies it. Anyone who lists this case as an "AI sanction" has added a fact the record does not contain. **That the sanction is final.** Counsel appealed to a district judge; the appeal was severed into *In Re: Appeal of Andrew Francisco*, No. 2:25-mc-00112 (E.D. Cal.). At a hearing on 7 July 2025 before Judge Dale A. Drozd, "Counsel further requested court vacate sanction order and remand matter to different magistrate judge," and the matter was taken under submission. The public docket we can reach shows no ruling; its last known filing is 1 November 2025 and no opinion has been published. As of 23 July 2026 we do not know whether the $1,500 still stands. ## Path to green There is no client to call — this is a court-origin record, and every figure above is already Tier 1 from a signed order, byte-tied to an Internet Archive capture by SHA-1. What stands between this entry and publication is one docket check: the Clerk's Office for the Eastern District of California (Sacramento), or a PACER pull on 2:25-mc-00112 (DAD), to establish whether the appeal has been decided. A ruling that vacates the sanction changes the story; a ruling that affirms it closes the last gap. Separately, whether either bar acted on the referral is a public-discipline-record question, not a court-file one.