Just when judicial ethics was getting boring again, one federal judge’s video dissent and a state supreme court’s new generative artificial intelligence (GenAI) avatar to report on their decisions have broken the mold on judicial speech.
A federal judge from the Ninth Circuit took to video to present his dissent from a hot gun rights decision. And the Arizona Supreme Court deployed AI-generated avatars as news readers to deliver hot news about their latest decisions.
Is any of this ethical? In your author’s view, at least some of it is.
Does any of it reflect good judging? Judge for yourself.
Recently, one lifetime appointee to the U.S. Court of Appeals for the Ninth Circuit dissented from a 6-4 en banc court decision upholding a controversial gun-control law.
At issue in this long-running litigation was the validity under the Second Amendment to the U.S. Constitution of California’s 2016 law criminalizing possession of large-capacity gun magazines. The decision in Duncan v. Bonta, decided March 20, 2025, centered on whether such magazines are “arms” and, if so, whether the Second Amendment permits their prohibition.
It took the Ninth Circuit 147 pages, five opinions and an 18-minute, 44-second video to decide.
Filmed in chambers, the video begins with dissenting Judge Lawrence VanDyke seated at a desk, American flag behind him and a rifle (Google says it’s probably an AK-47) mounted behind him. Judge VanDyke tells viewers he believed his 25-page written dissent needed a video demonstration to drive it home. As he says, “showing is much more effective than telling.”
After a short intro, Judge Van Dyke cuts to part of the oral argument video to show his colleagues and prevailing counsel’s fundamental lack of understanding basic firearm technology.
Next, we see the judge standing before a conference table with several guns, all rendered inoperable and safe, he stresses. (No explanation of how they got past the U.S. Marshals Service.) During his demo, he takes apart handguns, displays a personal firearm and shows a part from his “competition gun.” All to demonstrate why a large-capacity magazine is clearly an “arm” and not a mere accessory.
The judge prominently links to the video in his dissent. The video is not posted with the decision, and it appears not to be part of the case docket. It’s posted on the court’s YouTube channel.
At last report, no one has called for Judge Lawrence VanDyke to quit his day job to professionally generate viral videos. (Life tenure may have advantages over influencer status.)
For years now, some judges have adopted a writing style that jurists of generations past would find, well, indecorous and unbecoming. Short, punchy sentences. Contractions, colloquial expressions and slang. Comments bordering on snarky.
The trend encompasses more than a plain-English approach; they seem to write for an audience accustomed to social media pithiness. Holmes and Cardozo are not their role models, at least as to opinion writing. So maybe a jaunty video supplementing a written opinion is just a natural progression.
As Judge Van Dyke wrote, “It’s so easy to demonstrate the conceptual failings of the majority’s new test that even a caveman with just a video recorder and a firearm could do it.” (Is he the self-deprecating “caveman”? Are his colleagues less smart than cavemen?)
To be clear, some of Judge VanDyke’s colleagues were not having it.
Judge Marsha S. Berzon, in a concurrence joined by the entire majority, noted that the court’s rules do not allow a video dissent; opinions must be written. Further, she writes, Judge Van Dyke arrogated to himself the role of expert witness, offering and considering facts not in the record to support his position. And, they argued, he can’t play both roles—expert and judge. Of course, Judge VanDyke vigorously disagreed.
Judge VanDyke’s colleagues don’t quite say he’s acting unethically.
Still, it is clearly unethical conduct for a judge to consider, offer or rely on facts not in the record (ABA Model Rule of Judicial Conduct 2.9(C)). His colleagues point out that, had the judge actually acted as an expert below, he’d have been required to recuse from hearing the case on appeal (ABA Model Rule 2.11(A)). Judge VanDyke says he did no such thing of course, only using the video to demonstrate more vividly, with props, the way his (and the majority’s) analysis works (or doesn't).
In addition, while judges have been known to read their written opinions from the bench, judges are generally not permitted to comment publicly about their cases, certainly not while they are pending (ABA Model Rule 2.10(A)). Had the judge not included a link in his opinion, and claimed it was part of his opinion, would it have been problematic for him to simply do a TED Talk on his opinion? Historically, some judges have, in fact, made very similar public statements about legal doctrine.
As of this writing, no motions to recuse Judge VanDyke have been filed, and there are no reports of any judicial misconduct complaints against him.
In other news, virtually every U.S. high court has professional spokespeople. Among their jobs is sometimes explaining their decisions in plain(er) English than a court’s opinion.
Arizona is now not only the land of dramatic lawyer regulatory reform— alternative business structures, licensed paralegals and fee-sharing, oh, my!—but the home of the first supreme court in the country to deploy GenAI-produced avatars as news readers to explain their new decisions. Meet Daniel and Victoria.
When I say “avatar,” I mean video images that show what seem to be two attractive young people, looking like your ordinary network or local TV news anchors, reporting just like Wolf Blitzer or Bret Baier. Yes, they very much look real; no, they are not real people.
They were created for the court with the program Creatify. Beginning in March, video by one of them accompanies every new Arizona Supreme Court decision.
For now, they appear to be simply news readers, cutting production time of explainer videos from hours to about 30 minutes. The justice who authors an opinion drafts a news release, the wording of which is approved by the entire court. The justice then works with the court's communications team on script for the avatars. The court’s human spokespeople insist that the avatars aren't interpreting decisions.
Their demeanor is a bit less than perfectly human today, but human court spokespeople say they’re exploring different emotional deliveries, cadences and pronunciations as well as Spanish translations.
Career prospects for Victoria and Daniel are not clear. Reports are that they may work on access to justice and other public and civic information efforts.
Still, given the capabilities of GenAI, some of us expect great things of Daniel and Victoria. Why shouldn't the court harness an AI—they’ll have to supervise, of course, just as we lawyers do—to actually read a new decision, plus the briefs and oral argument, plus the decisions it cites, and generate a plain English summary. If AIs attached to Westlaw and Lexis can’t do that today, I’d be shocked.
Is using AI-generated avatars an ethical way for judges to speak? Undoubtedly. Just as a court can issue a summary press release along with a decision, neutrally explaining what the court decided and why, a court can hire a professional spokesperson to do so, as many do. So of course, it can hire an AI avatar to do the same thing. Had Judge VanDyke done merely that, his colleagues would have been hard pressed to criticize him.
Stay tuned to see whether judges begin using robed, AI-generated avatars to explain their decisions in Instagram Reels.
