ChatGPT fabricates information about precedents and embarrasses the lawyer using it

Steven Schwartz, an attorney at Levidow, Levidow & Oberman, may be subject to injunctive relief because he submitted information to the court that was later found to be untrue: it was simply invented by the ChatGPT chatbot powered by artificial intelligence, which the lawyer used.

As reported by the New York Times, the lawyer represented a client named Roberto Mata, who suffered a knee injury from a food cart during an Avianca flight. Attorney Steven Schwar decided to provide the judge with a list of cases where plaintiffs have sought compensation in such cases, but he enlisted the help of the ChatGPT chatbot to gather that information.

The ChatGPT chatbot, in response to a lawyer's query, provided several similar cases, such as "Martinez v. Delta Airlines" and "Durden v. KLM Royal Dutch Airlines". When Schwartz submitted all the necessary documents to the court, the Avianca airline team found that such precedents did not actually exist. it turned out that they were invented by a chatbot.

The lawyer claimed that "he was not aware that the content provided by the chatbot may not be correct. The judge has scheduled a hearing next month to discuss "possible sanctions" against the lawyer who entrusted his case to artificial intelligence.