Mata v. Avianca, Inc.
U.S. District Court, S.D.N.Y. (No. 22-cv-01461)
- Judge
- Hon. P. Kevin Castel
- Date filed
- 2022-02-22
- Sanctioned
- 2023-06-22
- AI tool
- ChatGPT (GPT-3.5)
- Monetary penalty
- $5,000 (joint and several — Steven A. Schwartz, Peter LoDuca, Levidow, Levidow & Oberman, P.A.)
- Outcome
- Rule 11 sanction; letter to each judge falsely identified as author of a fabricated opinion; letter to Mata explaining what happened; case ultimately dismissed on statute of limitations grounds.
What happened
Roberto Mata sued Avianca Airlines for a knee injury sustained on a 2019 flight from El Salvador to New York. When Avianca moved to dismiss on statute-of-limitations grounds under the Montreal Convention, plaintiff's counsel Steven Schwartz — a 30-year attorney at Levidow, Levidow & Oberman — filed an opposition brief containing citations to at least six judicial opinions that did not exist. Schwartz had generated the brief with ChatGPT and, when opposing counsel and the court could not locate the cited cases, submitted a second ChatGPT-generated affidavit with fabricated "screenshots" of the opinions.
Judge Castel ordered a show-cause hearing at which Schwartz admitted he had used ChatGPT for legal research, had asked the model whether the cases were real, and had accepted the model's reassurance that they were. Peter LoDuca, the attorney of record who signed the filings without independent verification, was sanctioned jointly. Judge Castel's June 22, 2023 opinion is a widely cited landmark on attorney duties when using generative AI: "Existing rules impose a gatekeeping role on attorneys to ensure the accuracy of their filings. [Schwartz and LoDuca] abandoned their responsibilities when they submitted non-existent judicial opinions with fake quotes and citations created by the artificial intelligence tool ChatGPT."
The court imposed a $5,000 penalty joint and several against Schwartz, LoDuca, and the firm, and required them to send letters to each of the real judges falsely identified as the author of a fabricated opinion. The court declined to impose harsher sanctions after finding the attorneys had eventually acknowledged the misconduct in good faith. The underlying case was dismissed on statute-of-limitations grounds.
The fabricated citations
Varghese v. China Southern Airlines Co., Ltd., 925 F.3d 1339 (11th Cir. 2019)
Does not exist. The real 925 F.3d 1339 is a different, unrelated case.
Shaboon v. Egyptair, 2013 IL App (1st) 111279-U (Ill. App. Ct. 2013)
Does not exist.
Petersen v. Iran Air, 905 F. Supp. 2d 121 (D.D.C. 2012)
Does not exist.
Martinez v. Delta Air Lines, Inc., 2019 WL 4639462 (Tex. App. Sept. 25, 2019)
Does not exist.
Estate of Durden v. KLM Royal Dutch Airlines, 2017 WL 2418825 (Ga. Ct. App. June 5, 2017)
Does not exist.
Miller v. United Airlines, Inc., 174 F.3d 366 (2d Cir. 1999)
The 174 F.3d 366 reporter cite is a real but unrelated case; the quoted holding attributed by ChatGPT was fabricated.
How Citation Safe would have caught this
Citation Safe checks every citation string in a filing against CourtListener's primary-source database of federal and state opinions before you file. In Mata, all six fabricated citations would have returned "cite not found in any authoritative reporter" verdicts within seconds — the exact "gatekeeping" step Judge Castel found the attorneys skipped. The same run would have flagged the quotations attributed to real-but-unrelated cases as "quote not present in the cited opinion." Total run time on a brief of this length: under 30 seconds.
Sources
Run Opposition Check on any brief you’re about to file for — $49.
Full annotated report on every citation in the opposing filing. Attach it to your Rule 11 or sanctions motion.