FOLLOW US

The advent of ChatGPT has revolutionized various industries, and the legal field is no exception. In a recent court case in New York, the use of ChatGPT has led to the creation of fake court citations. And has one attorney facing potential sanctions.

The lawsuit involved Roberto Mata, who claimed to have suffered injuries from a serving cart while traveling on Avianca Airlines in 2019. Steven Schwartz, an experienced attorney from Levidow, Levidow & Oberman, representing Mata, handled the case. 

Judge Kevin Castel of the Southern District of New York discovered that at least six cases submitted by Schwartz as part of his research for the case were suspicious.

According to Judge Castel's order, these cases seemed to be fabricated, complete with bogus judicial decisions, quotes, and internal citations.

The surprising revelation was that the source behind these fake cases was ChatGPT, an AI language model. Schwartz utilized ChatGPT to assist him in preparing the brief.

In April, Condon & Forsyth, Avianca's lawyers wrote a letter to Judge Castel questioning the authenticity of the cases Schwartz detailed in the affidavit.

The Judge noted the unprecedented situation in his May 4 order, stating, "The court is presented with an unprecedented circumstance." 

AI creates fake court citations 

Among the supposedly cited cases was Varghese v. China South Airlines, Martinez v. Delta Airlines, Shaboon v. EgyptAir, Petersen v. Iran Air, Miller v. United Airlines, and Estate of Durden v. KLM Royal Dutch Airlines. 

However, neither the judge nor the defense could verify the existence of these cases.

Schwartz, in an affidavit, claimed that he had never used ChatGPT for legal research prior to this case and was unaware of the possibility that its content could be false.

 However, he accepted responsibility for not verifying the sources provided by the chatbot. As a result, Schwartz is now facing a sanctions hearing on June 8.

Expressing remorse in a recent affidavit, Schwartz stated, "I greatly regret having utilized generative artificial intelligence to supplement the legal research performed herein and will never do so in the future without absolute verification of its authenticity."

 Additionally, fellow attorney Peter Loduca, in an affidavit, supported Schwartz's research and claimed he had no reason to doubt its sincerity or any involvement in the research process.

On April 25, Schwartz was ordered to show cause as to why he should not be sanctioned for using a false and fraudulent notarization in an affidavit.

 In a subsequent affidavit filed on Wednesday, Schwartz provided screenshots of his conversation with ChatGPT, attempting to confirm the authenticity of the cases.

Despite asking ChatGPT if "Varghese" was a real case, to which the chatbot affirmed it was. Schwartz later inquired about the source, and ChatGPT again maintained that the false case was genuine. 

The chatbot apologized for any confusion and later stated that the case could be found on legal research databases such as Westlaw and LexisNexis.

When Schwartz sought assurance that the other cited cases were also real, ChatGPT reassured him that they were genuine. And could be verified on reputable legal databases.

ChatGPT lied. And Schwartz is still uncertain what the consequences will be.