Sanctions and Warnings for Attorneys’ Use of ChatGPT

Emojis are pretty old technology, and yet, as we discussed last week, the law is just wrestling with whether assent and signature by emoji is effective, and so far, only Saskatchewan has weighed in on the issue.  The rest of the common-law world is sprinting to catch up.  

And so it is with the new AI chatbot technology.  Back in June, as Sara Merken reported in Reuters, two New York attorneys were disciplined for relying on ChatGPT to write a brief and not noticing that the bot had hallucinated cases.  Apparently, the lawyers failed to acknowledge the error in a timely way when opposing counsel and the court could not find the cases.   Although the court found that the attorneys had acted in bad faith and had made false and misleading statements, the court also expressed understanding that attorneys might make legitimate use of chatbots, subject to attorney gatekeeping “to ensure the accuracy of their filings.” 

Benjamin Weiser, writing in The New York Times, digs deeper, covering the sanctions hearing in detail.  The attorney who was led astray heard about ChatGPT from his college-aged children but had never worked with it before.  “I used this to write all my papers this semester, Dad, and look: 2.73 GPA!”  He never imagined that it would make up cases, and why would he?  Never trust technology recommended by anyone under thirty.  

Meanwhile, other courts are being proactive, as Jacqueline Thomsen reports for Reuters, U.S. District Judge Brantley Starr of the Northern District of Texas issued at order putting attorneys on notice “that they can’t just trust those databases. They’ve got to actually verify it themselves through a traditional database.”  Other courts require attorneys to disclose when they use AI in their research.  Eugene  Volokh reports on his beloved eponymous conspiracy blog that at least one judge has prohibited attorneys admitted pro hac vice from using AI.   

That seems a bit harsh.  ChatGPT is a tool like any other research tool.  Prohibiting use of ChatGPT makes as much as sense as prohibition of use of Google or Wikipedia.  Except that those two tools have been around for a while,  But then again, I am not worried that either of them will become sentient and destroy humanity.