Episode 025 Do Lawyers Need New Rules for AI? Judge Elizabeth Gunn Says ... Maybe Not

A federal bankruptcy judge just told us she doesn't personally use generative AI in her own chambers - and that's exactly why this conversation is worth your time. Judge Elizabeth L. Gunn, who sits on the U.S. Bankruptcy Court for the District of Columbia, spent nearly an hour with Ron and Heather working through what AI actually changes about the practice of law, and what it doesn't.

Central question: If AI can now do in seconds what used to take a junior associate hundreds of hours, who trains the next generation of lawyers - and who's accountable when the machine gets it wrong?

In this episode:
- Why Judge Gunn deliberately limits AI use in her own chambers, to protect it as a training ground for law clerks and interns
- How judicial opinions on AI vary widely across the bench - "as many opinions as there are black robes"
- Whether it matters, evidentiarily, if a summary took an associate 700 hours or an AI tool 700 seconds to produce
- How firms could develop internal AI tools for chapter 11 practice, and what foundation they'd need to lay for the court
- Why Judge Gunn believes generative AI is hollowing out the mid-level associate pipeline
- How fee applications are starting to raise questions about AI subscriptions and engagement agreements
- Judge Gunn's real experience with hallucinated cases in her courtroom, including sanctions and a running database of violations
- Why she believes a sliding scale of culpability makes sense for different kinds of hallucinations in legal briefs

We also discuss:
- Whether courts should look behind a flat-fee agreement to question consumer bankruptcy AI tool charges
- How Lexis and Westlaw brief-checker tools are used in chambers to catch bad citations
- Why Judge Gunn is skeptical the UAE's centralized, nationwide AI judiciary approach would work in the U.S. federal system
- Differing philosophies among law schools on AI in the classroom, from banning electronics to teaching practical AI skills
- Whether it's the court's role to inquire into AI-related terms of an engagement agreement when reviewing a fee application

Key Takeaway:

Judge Gunn's core message is that AI doesn't require a parallel set of rules. Rule 9011, the rules of evidence, and existing duties of competence and candor already cover it. What changes isn't the standard - it's how carefully lawyers have to apply it, and how quickly courts have to learn to tell the difference between an inaccurate pin cite and an argument built on a case that doesn't exist.

For Flintstones lawyers, that's reassurance: you don't need to master new technology to stay compliant, because your existing ethical obligations already cover AI-created work product. For Simpsons lawyers dabbling with tools like Claude or ChatGPT, Judge Gunn's evidentiary framework is a warning - if you can't establish the foundation for how a summary was created, speed doesn't save you. For Jetsons lawyers building AI into chapter 11 workflows, her comments point to exactly what a judge will want documented before she trusts the output.

Mentioned in This Episode:
- Judge Elizabeth L. Gunn, U.S. Bankruptcy Judge, District of Columbia
- Bad Boys of Bankruptcy podcast (Judge Gunn's podcast)
- Claude
- ChatGPT
- Harvey
- Lexis
- Westlaw
- Best Case (bankruptcy case management platform)
- Rule 9011
- Northern Pipeline Construction Co. v. Marathon Pipe Line Co.
- Stern v. Marshall
- Nancy Rappaport, UNLV School of Law
- University of Chicago School of Law
- New York and Florida AI ethics rules
- DC Bar ethics rule on AI

Highlight Reels on YouTube and Instagram


[email protected]