Judges Defend AI Use by Attorneys for Early Brief Drafts

Some judges back the use of generative artificial intelligence by lawyers as an early drafting tool amid concerns about use of the emerging technology for court submissions.

Speaking on an online panel Wednesday, Chief Magistrate Judge Helen Adams of the US District Court for the Southern District of Iowa said courts “shouldn’t assume everything about AI is nefarious and not helpful.”

“I don’t think this is any different than relying on an associate or an intern to draft the brief,” Adams said.

Courts are grappling with how and whether to respond to the advent of generative AI services like ChatGPT, that create content in response to user prompts.

Use of AI by lawyers has led to publicized incidents of errors submitted to courts, including a case in a Manhattan federal court when a lawyer submitted an AI-created brief with fake case citations in litigation against an airline.

The webinar, held by the nonprofit Sedona Conference ahead of a planned in-person conference next month in Virginia on AI and civil litigation, comes as courts attempt to address the matter prioritized by Chief Justice John Roberts.

He made the rise of AI a key feature of his year-end report released Dec. 31, where he predicted that judicial work “will be significantly affected by AI.”

‘A Tool’

Judge Xavier Rodriguez of the Western District of Texas, another panelist on Wednesday, said he’s more concerned about ways courts have reacted to the use of AI, though he clarified the technology should only be used for initial drafting.

“I have no problem at all with AI tools being used to draft a motion or a brief. An AI tool is just that: a tool,” Rodriguez said.

Rodriguez said any orders issued by courts or judges in response to incidents like the case in Manhattan should be “tool agnostic” and not name ChatGPT or another specific AI service.

Judge Scott Schlegel, a state court judge in Louisiana on the panel, also raised concerns about judicial orders that use broad language and fail to distinguish between generative AI, and AI more broadly, which could include basic spell check and grammar services.

“I think we do the bar a disservice when we do general orders like this,” Schlegel said.

The panelists also warned that attorneys using AI must not only verify information for accuracy, but also avoid inputting confidential information into free versions of those services, which learn from user prompts.

Circuits Respond

Circuit and other trial courts have responded to the rise of generative AI in various ways.

A representative for the California-based US Court of Appeals for the Ninth Circuit said in January the court is creating a committee to focus on the impact of artificial intelligence on legal practice.

The New Orleans-based Fifth Circuit, which covers Texas, Louisiana and Mississippi, has also proposed requiring lawyers to certify that they verified the accuracy of any materials created by an AI service before submitting it.

That proposal drew criticism from lawyers who said the rule was “unnecessary,” “ambiguous,” and could discourage lawyers from using AI in beneficial ways.

The Eastern District of Texas has also adopted a rule requiring lawyers to verify materials created by AI.

Judge Arun Subramanian of the Southern District of New York updated practice guidelines for his courtroom last year to state that while use of ChatGPT and other similar technologies is not prohibited, attorneys “must at all times personally confirm for themselves the accuracy of any research conducted by these means.”

Judge Michael Baylson of the Eastern District of Pennsylvania issued an order in June stating that if AI was used, lawyers must “disclose that AI has been used in any way in the preparation of the filing.”