In a surprising turn of events, Michael Cohen, Donald Trump's former personal lawyer and fixer, admitted to unwittingly passing along bogus artificial intelligence-generated legal case citations to his attorney before they were submitted to a judge. This revelation came to light in a court filing that was unsealed on Friday in Manhattan federal court. The admission came after Judge Jesse Furman had asked for an explanation regarding the inclusion of court rulings that did not exist in a motion submitted on Cohen's behalf.

The AI-generated cases were cited as part of written arguments made by attorney David M. Schwartz, who was attempting to bring an early end to Cohen's court supervision after serving over a year behind bars. Cohen had previously pleaded guilty in 2018 to tax evasion, campaign finance charges, and lying to Congress, claiming that Trump had directed him to arrange hush money payments to a porn actor and a former Playboy model to protect his 2016 presidential campaign. Cohen, who was disbarred five years ago, stated in a declaration submitted to the judge that he found the citations through research using Google Bard and was unaware that the service could generate nonexistent cases. He explained that he used the internet for research since he no longer had access to formal legal-research sources.

Google Bard, which was introduced earlier this year as a competitor to Microsoft's ChatGPT, allows users to generate text based on prompts. However, these tools have been known to create false information, also referred to as "hallucinations." Cohen placed the blame on his lawyer and longtime friend, Schwartz, for failing to verify the validity of the citations before submitting them to the judge. He requested that the judge show leniency towards Schwartz, describing his failure to check the citations as an "honest mistake" and "a product of inadvertence, not any intent to deceive."

Schwartz, in a declaration filed with the court, claimed that he believed drafts of the papers were reviewed by E. Danya Perry, a former federal prosecutor now in private practice who also represents Cohen. However, Perry stated that Schwartz's claim was incorrect, as she had no involvement in the back-and-forth with Schwartz or his paralegal. When Perry discovered the false case citations, she promptly reported them to the judge and federal prosecutors. She emphasized in her submission to the judge that Cohen had not engaged in any misconduct and should not suffer any consequences due to Schwartz's misstep.

This incident involving fake citations generated by artificial intelligence is not an isolated case. Earlier this year, two lawyers in an unrelated case were fined $5,000 for citing bogus cases that were invented by ChatGPT, an AI-powered chatbot. The judge presiding over Cohen's case noted the similarity and discussed possible sanctions for the lawyers involved.

In his 2018 guilty plea, Cohen did not disclose the names of the two women who received hush money or directly implicate Trump. However, the amounts and dates aligned with the payments made to porn actor Stormy Daniels and Playboy Playmate Karen McDougal, both of whom claimed to have had affairs with Trump. Trump, who has since pleaded not guilty to multiple criminal charges, including allegations of falsifying business records, has consistently denied these allegations and accused them of being politically motivated.

As the news of Cohen's unwitting use of AI-generated citations continues to unfold, it raises questions about the potential risks and implications of relying on artificial intelligence in the legal field. It serves as a reminder that technology, while powerful, can also produce misleading or false information if not properly vetted. The court will now have to determine the appropriate measures to address this incident and its impact on Cohen's case.