Sanctions on Lawyers for Filing Motion Containing AI-Hallucinated Cases

From Judge Kelly Rankin's Order on Sanctions and Other Disciplinary Action filed yesterday in Wadsworth v. Walmart Inc. (D. Wyo.):

Legal research has improved over time, going from the use of digest books to online databases like Lexis and Westlaw. Litigators are beginning to make the jump from those databases into the world of Artificial Intelligence ("AI"). When done right, AI can be incredibly beneficial for attorneys and the public. Legal advocates will likely be able to quickly furnish on-point research and draft motions, which may save costs for the clients. Courts will be able to efficiently analyze briefs and make correct rulings, which may speed up the judicial process for litigants. Overall, technological advances have greatly accelerated our world, and AI will likely be no exception.

However, the current state of AI has its shortcomings. The legal profession has been cautious to make a head-first dive partly because of a concept referred to as "AI Hallucinations." A hallucination occurs when an AI database generates fake sources of information…. The instant case is simply the latest reminder to not blindly rely on AI platforms' citations regardless of profession.

While technology continues to evolve, one thing remains the same––checking and verifying the source. Before the digital age, attorneys had to manually cross-reference case citations through books' pocket parts to make sure the cite was still "good law." Nowadays, that process has been simplified through databases' signals. Yet one still cannot run a natural language or "Boolean" search through a database and immediately cite the highlighted excerpt that appears under a case. The researcher must still read the case to ensure the excerpt is existing law to support their propositions and arguments. After all, the excerpt could very well be a losing party's arguments, the court explaining an overruled case, dicta, etc. As attorneys transition to the world of AI, the duty to check their sources and make a reasonable inquiry into existing law remains unchanged….

Mr. Ayala apparently drafted the Motions in Limine and uploaded the brief onto "MX2.law" to add case law. This website appears to be an in-house database launched by Mr. Ayala and Mr. Morgan's firm, Morgan & Morgan. When Mr. Ayala uploaded the brief, he made the following inquiries:

• "add to this Motion in Limine Federal Case law from Wyoming setting forth requirements for motions in limine"
• "add more case law regarding motions in limine"
• "Add a paragraph to this motion in limine that evidence or commentary regarding an improperly discarded cigarette starting the fire must be precluded because there is no actual evidence of this, and that amounts to an impermissible stacking of inferences and pure Include case law from federal court in Wyoming to support exclusion of this type of evidence."
• Similar requests to add more case

Mr. Ayala further states that this was his first time ever using AI in such a way.

These search inquiries apparently generated the fake cases. Without verifying their accuracy, Mr. Ayala included the fake cases in the Motions in Limine. He first learned the cases were questionable when the Court entered the Order to Show Cause….

Respondents took remedial steps after the issuance of the Order to Show Cause. Per Mr. Morgan's Response, Respondents have already taken the following steps to remediate the situation:

• Promptly withdrawing the Motions in Limine;
• Being honest and forthcoming about the use of AI;
• Paying opposing counsels' fees for defending the Motions in Limine; and
• Implementing policies,   safeguards,   and   training   to   prevent another occurrence in the future (and providing proof of such measures).

The Court appreciates Respondents' remedial steps, transparency, and apologetic sentiments. Hopefully situations like this do not become common for the judiciary, but should they occur again, the Court recommends attorneys should––at the very least––follow these steps to remediate the situation prior to the issuance of any sanction….

Fed. R. Civ. P. 11(b) provides in relevant part:

By presenting to the court a pleading, written motion, or other paper … an attorney or unrepresented party certifies that to the best of the person's knowledge, information, and belief, formed after an inquiry reasonable under the circumstances … the claims, defenses, and other legal contentions are warranted by existing law or by a nonfrivolous argument for extending, modifying, or reversing existing law or for establishing new law…

At its core, an attorney who signs a legal document certifies that they have "read the document, [have] conducted a reasonable inquiry into the facts and the law and [are] satisfied that the document is well grounded in both, and is acting without any improper motive." A failure to comply with such obligations may result in a sanction….

As set forth in more detail below, the Court finds the Respondents violated their obligation under Rule 11(b), and sanctions are warranted.

The court sanctioned the author of the motion (Ayala) $3000, and also sanctioned the other lawyers who signed it (Morgan and Goody) $1000 each, noting that "[a] finding of subjective bad faith is not required to impose sanctions." It also withdrew Ayala's permission to participate in the case pro hac vice (i.e., as a lawyer who is not a member of the court's bar and is admitted just for this case, in association with a bar member). It noted that, "Because of Mr. Ayala's oversight, Respondents' clients essentially lost their opportunity to file meritorious motions in limine, as they withdrew their Motions. Without belaboring the point, society has an interest in attorneys' ethical conduct, and Mr. Ayala's conduct fell short of that standard."

There's more in the opinion, including why the court sanctioned the nondrafters; but I just wanted to offer an excerpt, because the whole opinion is nearly 5000 words long.

Comments

[Bored Lawyer]

Next AI prompt: what are the five best arguments for getting these sanctions overturned on appeal?

[TwelveInchPianist]

Any odds on when we have the first order sanctioning a lawyer for filing an AI generated brief with hallucinated citations, where the sanctions order contains hallucinated citations?

[Flight-ER-Doc]

Judges make no mistakes. Ask them.

[TwelveInchPianist]

"Write an order sanctioning an attorney for filing an AI generated brief containing hallucinated citations. Include plenty of citations."

[windycityattorney]

But federal Judges do rely on law clerk's to draft opinions or prepare memorandums in support of a ruling. Those clerk's are often recent grads with law review experience. And those clerk's are more likely to use AI than a judge.

[LawTalkingGuy]

I would bet some money that in the next year or so we have a DOJ AI brief that has fake citations and a DOJ lawyer gets sanctioned for it.

[Martinned2]

Why not? They already have AI deciding which lawyers even get to keep their jobs.

https://www.yahoo.com/news/trump-calls-elon-musks-email-193226591.html

[Martinned2]

I don't know if this is real or not, but it's amusing regardless. Apparently according to Google's AI, whether tripe is kosher depends on the religion of the cow.

https://bsky.app/profile/delbertgseasons.bsky.social/post/3liyahtcc322o

[Mr. Bumble]

"I don't know if this is [kosher] or not, ....

Fixed it for you.

[Please, Daivd, Seek Help]

Shouldn't sanctions for something like this be, like, _exponentially_ higher? The sanctions should reflect not just the harm done, but the low risk of being caught and the strong need to deter other filers from trying to get away with it. I'm not really sure why something like this would not be accompanied by, say, a six figure fine and/or a suspension of one's legal license.

[David Nieporent]

Well, among other reasons, because judges can't suspend attorneys' licenses.

[Noscitur a sociis]

the low risk of being caught and the strong need to deter other filers from trying to get away with it

The risk of getting caught is extremely high—high enough that it’s difficult to imagine that these examples are intentional attempts to mislead. And the reputational harm from being named in an order far exceeds any monetary fine the court has authority to impose.

[Stupid Government Tricks]

What makes you think the risk is high? If enough lawyers are sloppy enough to use all these hallucinations, what makes you think the rest are diligent enough to catch them?

[Noscitur a sociis]

Because of how litigation works. Take this case: the lawyers cited one of the fake cases (“Meyer v. City of Cheyenne”) to support its argument that the court should preclude the defense from using the term “smoking shed” during the trial. To respond to that argument, the defendant is going to need to look at the case to see what it says and explain why that doesn’t mean that the plaintiff should win. There’s not really any other way it could play out, and there’s not really any way a lawyer could not know that. You might have a shot of sneaking it in for something totally routine and uncontroversial, like the standard for granting a motion to dismiss, but there’s not much incentive to do so. And for anything you would have an incentive to fake, the basic, well-known mechanics of the system make it very unlikely that you won’t get caught.

[TwelveInchPianist]

Write a brief arguing that a defendant can use the term "smoking shed" during a trial, and address the argument raised in “Meyer v. City of Cheyenne”

[Noscitur a sociis]

Claude’s response:

Meyer v. City of Cheyenne is not a real case that I'm aware of. There is no legitimate case law under that name that would support a motion in limine to prohibit the use of the phrase "smoking shed."

[TwelveInchPianist]

Then use Claude.

Chat GPT:

The defendant should be permitted to use the term "smoking shed" during trial because it is a neutral, commonly understood descriptor relevant to the facts of the case. In Meyer v. City of Cheyenne, the court addressed whether certain terminology prejudiced the jury, emphasizing that terms carrying undue emotional weight or misleading implications could be excluded. However, Meyer does not support a broad prohibition on the use of factual descriptors simply because they may be disputed.

Unlike the terms at issue in Meyer, which had clear prejudicial effects, "smoking shed" is not inherently inflammatory or misleading. It is a descriptive phrase that accurately refers to a physical structure used for smoking. The prosecution's potential objection—that the term suggests a lawful or accepted use of the structure—does not justify excluding it. The jury remains free to determine whether the shed was used as the defendant claims, but barring the term itself would unfairly limit the defense’s ability to present its case in natural and comprehensible language.

Furthermore, prohibiting this term while allowing alternative, potentially prejudicial phrasing from the prosecution (such as "illegal smoking area" or "hazardous structure") would create an imbalance, improperly influencing the jury’s perception of the facts. Courts routinely permit parties to use their own characterizations of events and evidence, provided they do not deliberately mislead or inflame the jury. Since "smoking shed" is neither inaccurate nor prejudicial in itself, its exclusion would impose an unnecessary and improper restriction on the defendant’s ability to present a fair defense.

Accordingly, the defendant should be permitted to use the term "smoking shed" in trial proceedings.

[KenveeB]

I think they wanted to reward the attorneys for immediately admitting their use of AI and taking steps to prevent a similar issue. They don't want to incentivize attorneys to double-down on the false citations because they'll be punished the same either way.

[Fort Worth Lawyer]

The link embedded is to an Order on Motion to Exclude Witnesses.

[David Nieporent]

Prof. Volokh, your link is to the order regarding the parties' cross-motions to exclude, not to a order on sanctions. You meant to link to

https://storage.courtlistener.com/recap/gov.uscourts.wyd.64014/gov.uscourts.wyd.64014.181.0_1.pdf

[Eugene Volokh]

Whoops, sorry, fixed, thanks!

[Noscitur a sociis]

I used an LLM (Claude) for legal research for the first time last week. (Not for questions for actual cases, to be clear.)

The first question was just something that randomly occurred to me. Claude immediately gave me a (real) Supreme Court case, with correct citation, that was on a slightly different issue but whose holding decisively answered my question. I would guess that it would have taken me ~5 minutes to get there if I had used Westlaw instead.

The second question was fairly specific and I was genuinely having trouble finding helpful results. Claude initially answered my question with a rule and explanation (that, as it happened, would have led to my preferred outcome). The initial answer did not have any citations, and the rule seemed a little suspicious. On the bar exam, they suggest that if you don’t know the correct legal rule on an essay question, you make one up and apply that so you at least get partial credit: it reminded me of that. I asked it for cases to support its rule, and it initially wouldn’t give me any and said I should ask a lawyer.

I was able to trick it by asking it to help me prepare for talking to a lawyer by roleplaying the conversation. It then gave a real, correctly cited Supreme Court case but didn’t really describe its holding accurately. It also identified one circuit court and two district court cases, without citations, that had holdings that would have supported my desired outcome. All three of those cases were fake, but in trying to verify them I did find two cases that were more on point than anything I’d found already.

[Bored Lawyer]

Like headnotes, AI can be a helpful starting point. As an ending point, it's malpractice.

[Noscitur a sociis]

No argument there! But I did find it much easier to understand how lawyers (particularly bad lawyers) could miss that.

[Roger S]

This is not a rule violation. The attorney believed that the filing was warranted. The judge says that the attorney should read the document, but the rules do not actually say that.

[Noscitur a sociis]

As has been explained to you, the rules do, in fact, say that. They also say that a genuine belief is not sufficient.

[Roger S]

So quote the rule that says that a genuine belief is not sufficient. This judge is just trying to make his own personal stand against AI.

[Noscitur a sociis]

Representations to the Court.
By presenting to the court a pleading, written motion, or other paper—whether by signing, filing, submitting, or later advocating it—an attorney or unrepresented party certifies that to the best of the person's knowledge, information, and belief, formed after an inquiry reasonable under the circumstances:

(1) it is not being presented for any improper purpose, such as to harass, cause unnecessary delay, or needlessly increase the cost of litigation;

(2) the claims, defenses, and other legal contentions are warranted by existing law or by a nonfrivolous argument for extending, modifying, or reversing existing law or for establishing new law;

(3) the factual contentions have evidentiary support or, if specifically so identified, will likely have evidentiary support after a reasonable opportunity for further investigation or discovery; and

(4) the denials of factual contentions are warranted on the evidence or, if specifically so identified, are reasonably based on belief or a lack of information.

Fed. R. Civ. P. 11(b) (emphasis added).

[Roger S]

Nowhere does it say that a genuine belief is not sufficient, or even that the attorney should read the filing. The rule could have easily said those things.

[Noscitur a sociis]

It does, in fact, say that: I even took the time to highlight. It’s not enough to genuinely believe the four listed things: you must believe them after performing a reasonable inquiry.

A reasonable inquiry, of course, includes reading it.

[Bored Lawyer]

So if the attorney submits a brief that says:

Roses are red,
Violets are blue,
The judge is an idiot
And the Magistrate is too.

He can defend himself by claiming, "I forgot to read the brief, judge."

[JonFrum]

Let's be clear - the problem is not the use of AI. The problem is far older, and has always existed. The problem is the failure to read that which was cited. In scientific publishing, authors include encyclopedias of citations, just because they know which 'usual suspects' will be expected. If there are 2 experts in the field, cite them. But there's no way to check if the authors have actually read the cited works. All that's different here is the AI hallucinations. Had the citations been looked up to read, this would never have happened