Sanction teardown · W.D. Kentucky, USA · 2025-11-10

United States v. Thomas Czartorski, et al.

What happened

In W.D. Kentucky, USA, a filing relied on ChatGPT to help draft legal argument. The court identified the following problems with the citations in that filing:

• Fabricated (Case Law)
  Wright's brief cites a purported opinion 'United States v. Hang Le-Thy Tran, No. 3:07-CR-53, 2008 WL 2699394 (E.D. Ky. July 3, 2008)' which the court identified as non-existent.
• Fabricated (Case Law)
  Brief cites 'United States v. Cope, 312 F. Supp. 2d 791 (E.D. Ky. 2004)' which the court treated as a fictitious opinion.
• Fabricated (Case Law)
  Wright cites 'United States v. Abbott, 2023 WL 4106534 (E.D. Ky. June 27, 2023)'; the court found the citation non-existent and noted internal inconsistencies.
• Misrepresented (Case Law)
  The brief quotes and characterizes United States v. Chavis, 296 F.3d 450 (6th Cir. 2002) in a manner the court found to misrepresent that Sixth Circuit holding.
• Misrepresented (Case Law)
  Wright cites United States v. Tran, 433 F.3d 472, 478 (6th Cir. 2006) but the court found the holding as presented was misstated.

Which AI tool

ChatGPT. Note: Charlotin's public database records tool attribution only where a court order, brief, or reporting on the matter states it explicitly; "unidentified" or "implied" means the record indicates AI use but does not name a specific product — we do not guess.

Outcome

Order to show Cause

Additional detail

In his response, Counsel acknowledged that he first researched relevant cases, and then "entered the cases into ChatGPT and requested that it highlight favorable arguments contained in the list of cases."

How Citation Safe would have caught this

Citation Safe runs three deterministic layers before a brief is filed: (1) does the citation exist against CourtListener's database of published opinions, (2) if quoted, does that exact language appear in the source, (3) does the cited case actually support the proposition it is cited for. Fabricated case citations fail Layer 1. Fabricated or misattributed quotations fail Layer 2 even when the underlying case is real. Misrepresented holdings — a real case cited for a proposition it does not support — are the target of Layer 3. None of these checks involve asking another language model whether the citation looks right; they are lookups and text-matches against the actual source, which is why a hallucinated citation has to survive a direct lookup against the authoritative source — not another model's opinion — to earn a VERIFIED stamp; our measured false-verify rate is published live at /quality.

Check a brief before you file it → · See our live false-verify rate

Source: https://www.damiencharlotin.com/documents/1024/USA_v_Czartorski_USA_10_November_2025.pdf, via Damien Charlotin's public AI Hallucination Cases Database (CC0).