I'm surprised -- very surprised, actually -- that the Court of Appeal only imposes $1,750 in sanctions against the attorney here. I would have imposed a greater penalty. As well as have been much meaner.
It's another one of these "AI hallucination" cases, where someone writes a brief in whole or part with artificial intelligence that contains made-up cases (or cases with made-up citations) that fail to support anything relevant. Once caught, what attorneys generally do in these settings is to deliver to the court a huge mea culpa -- sometimes combined with a "Oh but we try real hard to catch these things, so sorry we failed in this particular case" defense. They generally end up getting sanctioned, of course, but the court at least appreciates their candor. And, in most cases, anyway, their apology. (Though, at the same time, also understanding the strategic nature of those admissions.)
The attorney here, by contrast, adopts a different approach. Once caught, after the Court of Appeal issued an order to show cause, he initially told the justices that the miscited cases were entirely his fault, and said that "the core allegation as to using 'AI hallucination' cases is entirely untrue" -- asserting instead that "none of the cases was the product of an AI hallucination, and these errors are entirely clerical in nature." He proposed that he be sanctioned $500 and not required to show up at the hearing on the OSC, but the Court of Appeal refused to discharge the writ.
At the hearing, the lawyer -- Jeffrey Dean Grotke -- admitted that he had used AI in preparing the brief, ostensibly only to better organize the brief. The Court of Appeal was not persuaded, and repeatedly held that Grotke's responses to the court were "not credible" (as well as lacked "candor") -- and described in substantial detail the basis for the court's belief in this regard. In the end, it sanctions Grotke $1,750, to be paid to the Court of Appeal, and directs that the opinion be sent to the State Bar.
All that seems entirely appropriate to me. It's good to publish opinions that make clear, even though it's crystal clear already, that attorneys have to be diligent in verifying their AI-generated work. And when they're not, it's entirely appropriate to not only publish the underlying opinion, as well as to refer to the sanctioned attorney by name, but to also send that opinion to the State Bar.
But I would have been even harsher here. It's one thing to prepare an AI-generated brief without checking the resulting work product. That's wrong, but it's only negligence. It's another thing, after being caught, to make representations to the court about that brief that are factually untrue. That's intentionally deceptive -- as well as premeditated.
Words like "not credible" and "lacks candor" are, in my view, entirely accurate to describe such conduct, but too soft. To me, that's an even more serious offense. And deserving of an even harsher response.
When you make a mistake, honorable people own up to it. To do otherwise, especially to a court, bespeaks volumes about one's character.
