David Kluft

#LegalEthics Tidbit: Can my #AI hallucinations be used against me in an unrelated criminal proceeding?

An NY attorney submitted some #AI hallucinated citations to a court, and the Court issued an order finding that his response to a show cause order was not entirely credible. Months later, the attorney was testifying as the defendant before a grand jury in an entirely unrelated criminal matter (he was accused of hitting a process ... (cont.)

https://www.nycourts.gov/reporter/current/3dseries/2026/2026_50842.shtml
#law #lawfedi

12:09pmWeb
Show threadDavid Kluft14h ago

... server with his car). The prosecution questioned the lawyer about the fact that a judge found his #AI explanation not credible, and also about the underlying fake citations.

On appeal, the Supreme Court held that this was improper. While “questions regarding the Federal court's finding that the defendant was not entirely credible was an appropriate cross-examination subject ... it was improper … to question the defendant on the underlying facts regarding use of A.I., ... (cont.)

Show threadDavid Kluft14h ago

... when that information itself was irrelevant to the defendant's credibility.” The prosecution instead “should have simply asked if the Federal judge had sanctioned the defendant after finding that his responses to the legal issue were not credible and presented admissible evidence to complete the impeachment if the defendant denied the finding and sanction.”

Because of this error and other issues, the indictment was dismissed with leave to re-present.