This time, it's the 11th Circuit.
The court's opinion reads like something that should be satire but unfortunately isn't. Counsel submitted an opening brief citing at least eight nonexistent cases. After opposing counsel pointed out the problem, he tried to fix it by withdrawing the bad citations. Except the eight cases he "withdrew" weren't the same eight fake cases from his opening brief.
Worse still?
The replacement list consisted of eight more hallucinated cases.
You almost couldn't script it better.
The 11th Circuit responded exactly as it should have. It reminded lawyers that competence requires "legal knowledge, skill, thoroughness and preparation." It emphasized that AI is no substitute for actual intelligence. And it referred the attorney for disciplinary proceedings.
Good.
But here's what bothers me about the inevitable reaction to opinions like this.
People will say, "See? AI is dangerous."
No. That's the wrong lesson.
AI didn't file the brief.
The lawyer did.
AI is an extraordinary tool. But it never replaces my judgment. Nor should it.
I would never file a brief without manually checking every citation and every quotation. I don't care whether those citations came from ChatGPT, Lexis, Westlaw, a first-year associate, or my own memory.
Verification is the job.
In fact, this isn't really an AI problem at all. It's the same professional obligation lawyers have always had.
When Lexis or Westlaw gives me a case, I Shepardize it before relying on it. I confirm that the quotation actually says what I claim it says. I make sure the proposition I'm citing is still good law.
That's not because I distrust Lexis or Westlaw. It's because I'm the one signing the brief. AI deserves exactly the same treatment.
The lawyers getting sanctioned aren't victims of unreliable technology. They're victims of their own willingness to skip the last—and most important—step in the process. They accepted the first answer a machine gave them without exercising independent professional judgment.
That's not innovation. That's laziness.
Every time another sanctions opinion like this gets published, it reinforces the false narrative that AI is incompatible with competent lawyering.
The opposite is true.
Lawyers who learn how to use AI responsibly will have a significant advantage over those who don't. They'll work more efficiently, spot more issues, and deliver better value to clients.
But only if they remember one simple rule: AI can draft your brief. It cannot sign your name.
And that's why the responsibility—for every citation, every quotation, every argument, and every word filed with a court—will always belong to the lawyer, not the software.