chevron-down Created with Sketch Beta.

Litigation News

Litigation News | 2023

Use of ChatGPT for Research Leads to Bogus Cases, Sanctions

Onika K. Williams


  • A federal district court sanctioned a law firm and two attorneys who used fake case law generated by ChatGPT.
  • ChatGPT is an artificial intelligence–based computer program.
  • When questioned by the court and opposing counsel, they continued to stand by the fabricated opinions.
Use of ChatGPT for Research Leads to Bogus Cases, Sanctions
XH4D via Getty

Jump to:

Citing nonexistent case law in court filings and repeatedly lying about it to the judge will lead to sanctions. A federal district court sanctioned a law firm and two attorneys who used fake case law generated by ChatGPT, an artificial intelligence-based computer program, and then continued to stand by the fabricated opinions even after opposing counsel and the court questioned the cases. ABA Litigation Section leaders see this case as a reminder that technology can be a useful tool, but attorneys are ultimately responsible for the accuracy of their legal work.

Unfamiliarity of Practice Area Leads to Questionable Behavior

The conflict in Mata v. Avianca, Inc. began when an airplane traveler alleged his leg was injured during a flight from El Salvador to New York City. The traveler sued the airline in the Supreme Court of the State of New York, New York County, and the airline removed the suit to the U.S. District Court for the Southern District of New York under the Montreal Convention.

Initially, a specific attorney was listed on the state court complaint. This attorney performed the substantive legal drafting. After removal to the federal court, a different attorney became attorney-of-record but did no substantive legal work. The two attorneys had practiced together for more than 25 years in New York state courts.

Before the federal court, the airline moved to dismiss arguing the case was time barred. The plaintiff’s attorney-of-record filed an opposition to the motion to dismiss with citations to several purported judicial decisions. Although the attorney-of-record signed and filed the opposition, it was researched and written by the drafting attorney. The attorney-of-record did not review any of the cited judicial opinions.

Attempts to Hide Mistake Leads to Unprecedented Sanctions   

In reply, the airline stated it was unable to find most of the cases cited in plaintiff’s opposition. The court conducted its own search and was also unable to find many of the cited authorities. The court asked the attorney-of-record to file an affidavit with copies of the seemingly nonexistent cases. In response, the attorney-of-record requested another extension to reply to the court by falsely claiming he was on vacation. Subsequently, the attorney-of-record executed an affidavit written by the drafting attorney with purported excerpts of all but one of the requested cases.

After further inquiry from the court, the plaintiff’s attorneys ultimately revealed that they relied on legal opinions provided by ChatGPT. The drafting attorney stated he had never used the technology before, did not realize that ChatGPT could create false content, and used ChatGPT because the firm had limited access to federal cases. The attorney-of-record did not review the legal pleadings before filing.

Subsequently, the court held an order to show cause hearing, fined the law firm and attorneys $5,000, and required them to apologize to each judge falsely identified as an author of the fake opinions cited in the opposition. The court explained that the attorneys violated Federal Rule of Civil Procedure 11 when they acted with subjective bad faith, the “knowing and intentional submission of a false statement of fact.” For example, the drafting attorney first stated that he only used ChatGPT to supplement his research but later admitted at the hearing that it was the only tool he used. The court explained that the attorney-of-record violated Rule 11 by not reading any of the cases cited in the opposition and lying about being on vacation when the court asked for additional information. The court concluded that while the use of technology is commonplace, “existing rules impose a gatekeeping role on attorneys to ensure the accuracy of their filings.”

Use Caution when Using Artificial Intelligence (AI) for Legal Research

Litigation Section leaders observe that this case is a harsh reminder that lawyers are ultimately responsible for the content of their court filings. “Oversight and manual review of the technology’s output is critical in strategic use of AI in litigation,” states Tiffany Williams Brewer, Washington, DC, Vice Chair of the Litigation Section. “When a lawyer is relying on computer generated research, the lawyer is still responsible for their own research,” agrees Michael D. Steger, New York, NY, cochair of the Section’s Solo & Small Firm Committee.

A lawyer’s ethical responsibilities do not end with the use of technology. “The use of AI in practice always implicates a lawyer’s fundamental dual ethical duties to provide competent and diligent representation to their clients, under their state’s corollary to the ABA Model Rules of Professional Conduct 1.1 (Competence) and 1.3 (Diligence),” observes Williams Brewer.

 As in this case, the use of AI “can also implicate the duty of candor that lawyers have as officers of the court in communicating with the court under Model Rule 3.3 (Candor Toward the Tribunal),” continues Williams Brewer.

In response to the increased use of AI in the legal practice, several courts have issued standing orders that range from the disclosure of the use of programs like ChatGPT to a reminder that lawyers are responsible for providing accurate statements when using AI. “One of the benefits of artificial intelligence is that it can provide some initial research to help guide the attorney on where to look next,” counsels Steger. “But, at this point, AI is not a panacea for legal research and is not going to provide a shortcut that the attorney can use to complete all of their research,” he warns.