US | United States District Court for the Southern District of New York
Docket: 1:22-cv-01461-PKC • first instance
2023-06-22
Mata v. Avianca, Inc. — opinion and order on sanctions
ChatGPT no human verificationfake case citationfake quotationcontinued advocacy after challengemisleading statements to the court
I. Executive Summary
The court found that counsel filed submissions containing non-existent judicial opinions with fake quotations and citations generated by ChatGPT and then continued to stand by those authorities after their authenticity was challenged.
It emphasized that using an AI tool is not inherently improper, but attorneys must verify authorities before filing. The court imposed Rule 11 (or inherent-authority) sanctions to deter repetition.
II. Conduct Analysis
Counsel used ChatGPT to generate case law support, copied the resulting citations and quotations into filings without verifying that the cited decisions existed or supported the stated propositions, and continued to advocate those authorities after the court and opposing counsel questioned them.
III. Legal Foundations
Fed. R. Civ. P. 11(b)
Fed. R. Civ. P. 11(c)
Court inherent authority
IV. Key Facts
1) The opposition submission cited and quoted multiple purported decisions that the opposing party and the court could not locate.
2) After the court ordered counsel to produce copies of the cited authorities, counsel submitted additional materials continuing to rely on the fabricated opinions.
3) Counsel ultimately acknowledged that ChatGPT was used and that the cited cases were fabricated (hallucinated) rather than real precedent.
4) After a sanctions hearing, the court imposed Rule 11 (or inherent-authority) sanctions focused on deterrence.
V. Consequences & Sanction
1) Sanctions were imposed for filing AI-generated fake authorities.
2) The lawyers were required to notify their client and the judges whose names were falsely invoked.
3) A monetary penalty was assessed and payable to the court registry within a set deadline.