A lawyer who adopted a non-existent past precedent made up by ChatGPT is ordered to pay $ 5000



A federal judge ordered a lawyer to pay $ 5,000 (about 720,000 yen) in a case in which a passenger's lawyer sued an airline about a dispute that occurred on an airplane submitted a ``non-existent past precedent output by ChatGPT'' has dismissed the original lawsuit against the airline.

Lawyers have a real bad day in court after citing fake cases made up by ChatGPT | Ars Technica

https://arstechnica.com/tech-policy/2023/06/lawyers-have-real-bad-day-in-court-after-citing-fake-cases-made-up-by-chatgpt/



Lawyers found $5,000 for using ChatGPT in brief | Fortune

https://fortune.com/2023/06/23/lawyers-fined-filing-chatgpt-hallucinations-in-court/

Passenger Robert Mata sued Avianca Airlines in 2022, claiming that a catering cart hit his knee in August 2019 while on a flight with Colombian airline Avianca Airlines.

The case was originally heard in New York state court, but was later transferred to the United States District Court for the Southern District of New York. Attorney Stephen Schwartz of LEVIDOW, LEVIDOW & OBERMAN, who initially handled the case in state court, was not qualified to practice law in federal court. was given.

Mr. Schwartz, who was in charge of document preparation even after the transfer, will use ChatGPT, a sentence generation AI, to create an opinion that quotes the ``nonexistent past precedent'' output by ChatGPT as it is. Mr. Loduca, who received this opinion, submitted it to the court in his own name without scrutinizing the contents.

It turned out that past cases that did not exist were made up in the application form created by the lawyer using ChatGPT - GIGAZINE



The court and the defendant's lawyers, who saw the written opinion describing the non-existent incident, pointed out that the content may be false, and requested Mr. Loduca and others to reconfirm the content. In response to this request, Mr. Loduca and others submitted a ``court opinion detailing the precedent described in the opinion'', but it turned out that this view was also a false view output by ChatGPT.

In the wake of the case, U.S. District Judge Kevin Castell accused Schwartz and Loduca of 'harming clients, wasting court time, and wasting the other party's time and money.' I ordered a total of $5000 for the two of us.

Since it was found that the written opinion submitted by Mr. Loduca et al. Judge Castel agreed with the defendant's argument that ``under the Montreal Convention, which provides for a two-year statute of limitations, plaintiffs have filed too late.''



In a later affidavit, Schwartz said, ``Until this incident, I had never used ChatGPT as a source of information for conducting legal investigations, and therefore about the possibility that its content was false. I didn't know about it.' 'I heard that there is a new site called ChatGPT, which is like a super search engine, and I used it.' On the other hand, Mr. Loduca claims that 'I believed that the document created by Mr. Schwartz, a colleague of 25 years, was reliable.'

In ordering the two lawyers to pay, Judge Castel said, 'This is a deterrent, not a penalty or reparation, because good lawyers are properly assisted by junior lawyers, law students, databases, etc.' AI is a new addition to these means.” “But all documents generated by AI must be checked for accuracy. We trust the sincerity of the defendants who made their statements.'

In addition to the payment of $ 5,000, Mr. Loduca and others are required to submit a report to six real judges who have been arbitrarily quoted in fake court cases. Judge Castel said, ``If you force an apology, it will not be a sincere apology. emphasized.



in Software, Posted by log1p_kr