Former Trump fixer Michael Cohen said in a court filing on Friday that he emailed his attorney false citations created by artificial intelligence by mistake, which the lawyer then filed with the court.

According to Cohen, who represented the former president as an attorney, he mistook the AI bot Google Bard for a “super-charged search engine” when looking up court rulings that would establish precedent for doing away with his supervised release.

He said in a filing that was first cited by The New York Times that the cases generated by the artificial intelligence program were not real.

2018 saw the sentencing of Cohen to three years in jail and three years of post-release supervision for his crimes of lying to Congress and making covert payments to women who claimed to have had romances with Trump.and neglecting to disclose earnings.

“Not keeping up with emerging trends (and related risks) in legal technology and not realizing that Google Bard was a generative text service that, like Chat-GPT, could show citations and descriptions that looked real but actually were not,” stated Cohen in the petition on Friday.

He continued, saying he expected his lawyer to “vet my suggested additions before incorporating them” and that he was unaware that the AI service could produce false cases. Cohen stated that from July 2022, he has been represented in the post-release supervision case by lawyer David Schwartz.

As he had every right to, he trusted his attorney. Cohen is being represented by E. Danya Perry in support of his move for an early termination of supervised release. “Unfortunately, his lawyer seems to have made an honest mistake in not verifying the citations in the brief he drafted and filed,” Perry said in a statement to NBC News. The court documents, she said, “show that Mr. Cohen did absolutely nothing wrong.”

In a correspondence with Perry, a district judge, contended on Thursday that Cohen should not be held accountable for Schwartz’s bogus citations in his submission.