On March 26, during an employment dispute hearing, DeWaard was sharply reprimanded for not informing the judge of the artificial intelligence during his oral arguments. Although the court allowed DeWald to submit the video, Judge Sallie Manzanet-Daniels was confused when an unknown speaker who was clearly not DeWald appeared on the screen.
"Wait a minute," Manzanette-Daniels interrupted the video as soon as the avatar finished its first sentence. "Is he the defense attorney in this case?"
"I created it," DeWald replied, "it's not a real person."
The avatar - a "tall, handsome guy" named Jim - was provided by an artificial intelligence avatar company called Tavus. Dewald said the video was submitted because he had difficulty speaking, but the court was not aware that its content was artificially generated.
"It would have been nice if you had known that when you applied. You didn't tell me, sir, and I don't like to be misled," Manzanette-Daniels said in response to Dewald's admission. "You don't think of this court as the beginning of your business."
It's the latest in a series of snafus that have occurred when people try to combine legal processes with artificial intelligence technology. In 2023, two attorneys and a law firm were punished for submitting false legal research fabricated by ChatGPT. "Robo-lawyer" company DoNotPay was also ordered to pay a $193,000 settlement to the Federal Trade Commission in February for promoting, without evidence, that its artificial intelligence legal representation services were as good as real lawyers.