Part of the AI Regulation News hub.
A published Virginia Court of Appeals opinion found that 25 of the 26 direct quotations in a brief were invention, and ordered the lawyer who signed it to complete a continuing legal education course on AI or appellate practice
The court counted them. Twenty-six purported direct quotations from Virginia cases, not one an exact match, one off by a single word and the remaining twenty-five pure invention. The sanction is unusual, and so is the reasoning about why a fine was not the right one.
Bottom line: A published opinion. The court held it was compelled by statute and precedent to impose a consequence, and required appellee's counsel to complete a continuing legal education course relating to AI or appellate practice within six months of the opinion. No monetary sanction was imposed.
Who this affects: Virginia litigators and anyone signing a brief in a Virginia court, appellate practitioners generally, and firms deciding what their sanction exposure looks like when assisted research goes wrong and is caught early.
Issue date: Opinion issued 25 August 2026. The CLE course must be completed within six months of issuance.
What changed: No new rule. The court applies Code section 8.01-271.1, under which a signature certifies that the paper is well grounded in fact after reasonable inquiry, and treats a brief full of invented quotations as engaging that certification.
Analysis: The mitigation analysis is the transferable part. The court records that the unreliable quotations were identified before oral argument, that no monetary losses were claimed or established, that opposing counsel never raised the issue, and that counsel took responsibility at oral argument. Those four facts, not the absence of fault, are why the sanction is educational rather than financial.
Primary sources: Published opinion, Court of Appeals of Virginia, Record No. 1407-25-4
- Case
- LPC Commercial Services, LLC, f/k/a LPC Commercial Services, Inc. v. Ar-Razzaaq Properties, LLC
- Record number
- Record No. 1407-25-4
- Court
- Court of Appeals of Virginia, argued at Alexandria
- Panel
- Judges AtLee and Friedman and Senior Judge Annunziata; published opinion by Judge Frank K. Friedman
- Below
- Circuit Court of Fairfax County, Patrick M. Blanch, Judge
- Representation
- Morgan P. Fryar, with Edward W. Cameron, Cameron Ingersoll Roche PLLC, for appellant. Michael Hadeed, Jr., Hadeed Law Group, P.C., for appellee.
- Jurisdiction
- Virginia, United States
- Provision applied
- Virginia Code section 8.01-271.1, with Rules of Professional Conduct 1.1, 1.3, 3.1 and 3.3 cited
- Sanction imposed
- Counsel for appellee to complete a continuing legal education course relating to AI or appellate practice within six months of issuance of the opinion. No monetary sanction.
- Disposition
- Affirmed, including the damages award under the right result for the wrong reason doctrine.
- Editorial Note
- Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.
- Primary source
- https://www.vacourts.gov/opinions/opncavwp/1407254.pdf
The count
The underlying appeal was a commercial real estate dispute about a Property Management Agreement, and the court affirmed. The part that will be read is Section III.
The court's own arithmetic is precise and worth reproducing exactly, because secondary accounts of these cases usually round it. Appellee's brief contained 26 purportedly direct quotes to Virginia case law. None of the alleged quotes was an exact match. One differed from the source material by only a word. The other 25 appeared to be pure invention.
Section III then describes what those 25 phantom quotes actually were, on a spectrum: general paraphrases of the case's actual language, optimistic or wishful paraphrases of the case's content, and totally hallucinated and presumably computer-fabricated quotes having nothing to do with the cited case.
That taxonomy is useful in itself. It distinguishes the paraphrase-dressed-as-quotation problem, which predates generative AI, from wholesale fabrication, and the court found both in a single brief.
How the court reasoned to a non-monetary sanction
Virginia Code section 8.01-271.1(B) provides that an attorney's signature certifies that they have read the paper and that, to the best of their knowledge, information and belief formed after reasonable inquiry, it is well grounded in fact and warranted by existing law. Subsection (D) provides that where a paper is signed in violation, the court shall impose an appropriate sanction, which may include payment of the other party's reasonable expenses and attorney fees.
The court sets out the range available: monetary sanctions, revocation of pro hac vice admission, disciplinary action, and striking non-complying documents, alongside the courts' inherent authority to maintain the integrity of judicial proceedings.
It then works through the facts that pulled against a financial penalty. The unreliable quotations were identified before oral argument, so the misrepresentations were prevented from sullying Virginia precedent. Counsel acknowledged the erroneous quotations, assured the court the misquotes were unintentional and that there was no intent to mislead anyone, and the court says it believes him. No monetary losses were claimed or established, and opposing counsel never raised the issue with the court.
The court's conclusion is that the source of the phantom quotes was some form of computer-assisted research that went awry, and that the brief aptly illustrates the perils of relying on assisted research that is not carefully checked.
It also says plainly that it would have preferred to leave it there: being the subject of this cautionary tale is, in the court's view, penalty enough at this juncture, but the court is compelled by statute and case law to impose a consequence. Hence the CLE requirement.
The passage lawyers will quote
Having found the violation, the court sets out why the responsibility cannot move to the tool. It begins sympathetically: the practice of law is often stressful, difficult and time-consuming, there are limited hours in a day, technology that lessens the load will present itself attractively, and there will be a strong gravitational pull on lawyers to rely heavily on time-saving technology.
Then the pivot. At the end of the day it is the attorney who signs the brief and is therefore responsible for the accuracy of its contents, under Code section 8.01-271.1 and Rules of Professional Conduct 1.1, 1.3, 3.1 and 3.3 on competence, diligence, non-frivolous claims and candour toward the tribunal.
The line that will travel is the court's mnemonic: to avoid a potential Code section 8.01-271.1 reckoning in this context, it is important for lawyers to remember that the A in AI stands for artificial. The technology, at least for now, cannot think for itself or exercise caution or responsibility, which makes it incumbent on legal professionals using AI to exercise human oversight and control over their research and filings.
The court then quotes an article from Virginia Lawyer for the proposition that AI is not designed to say I don't know, that these tools rely on predictive language models producing plausible responses rather than reasoned judgment, and that competence today requires vigilance, verification and skepticism, especially when technology offers answers with confidence but without understanding.
A second development, flagged in a footnote
The opinion carries a footnote that is a story in itself for Virginia practitioners. The court records that the Virginia State Bar recently introduced, for comment, Legal Ethics Opinion 1902, dealing with opposing counsel's responsibilities when confronted with a brief fraught with AI-hallucinated misstatements of law or fact.
That matters here because of the court's own observation that appellant's counsel never raised the phantom quotes with the court. Under the existing framework that omission counted, if anything, in favour of leniency toward the offending side, since it meant nobody had incurred expense contesting the point.
A rule addressing what opposing counsel must do on encountering hallucinated authority would change that calculus, converting a discretionary courtesy into a duty and creating exposure on both sides of a brief rather than only the signing side.
The opinion does not decide anything about LEO 1902, which was out for comment. It is flagged here because the two developments read together are the direction of travel in Virginia.
What to take from the outcome
The headline is not the leniency. It is that the court treated a sanction as mandatory. Virginia precedent, the opinion says, calls for a sanction against one who files a brief violating section 8.01-271.1, and the court considered itself compelled to impose a consequence even while suggesting that the public cautionary tale was punishment enough.
The discretion, therefore, sat in the choice of sanction rather than in whether to sanction. A CLE requirement on AI or appellate practice, completed within six months, is at the mild end of a range that runs through monetary sanctions and revocation of pro hac vice admission to disciplinary referral.
For a practitioner assessing exposure, the four mitigating facts are the checklist: was the problem caught before argument, was there any monetary loss, did the other side incur expense pursuing it, and did counsel take responsibility promptly and unequivocally when confronted.
The court's closing thought is unusually candid, and applies to the bench as much as the bar: given the pace at which artificial intelligence is advancing, we would all be well-advised to closely monitor its expanding capabilities, and also its limitations.
Three practical points. First, the sanction was not discretionary in principle: the court read Virginia precedent as requiring a consequence for a brief that violates Code section 8.01-271.1, and exercised its discretion only over which one, landing on a CLE course on AI or appellate practice within six months. Second, the mitigation that produced that outcome is a specific and reproducible set of facts, not a general finding of good faith: caught before oral argument, no monetary loss claimed or established, not raised by the opponent, and responsibility taken at argument. Third, watch Legal Ethics Opinion 1902, out for comment at the Virginia State Bar, which addresses what opposing counsel must do when faced with a brief containing AI-hallucinated authority - because in this case the opponent's silence is part of why the sanction was light.
Source File
https://www.vacourts.gov/opinions/opncavwp/1407254.pdf
Open the opinion and confirm four things: the caption showing Record No. 1407-25-4 with the panel of Judges AtLee and Friedman and Senior Judge Annunziata and the issue date of 25 August 2026; footnote 6, recording 26 purported direct quotes of which 25 appear to be pure invention; Section III, setting out Code section 8.01-271.1 and the phrase that the A in AI stands for artificial; and the end of Section III imposing the continuing legal education requirement within six months.
it is important for lawyers to remember that the "A" in "AI" stands for artificial. This burgeoning technology, at least for now, cannot think for itself or exercise caution or responsibility. LPC Commercial Services, LLC v. Ar-Razzaaq Properties, LLC, Record No. 1407-25-4 (Va. Ct. App. 25 August 2026)
FAQ
How many quotes were wrong?
The brief contained 26 purportedly direct quotes to Virginia case law. None was an exact match. One differed from the source by a single word, and the other 25 appeared to be pure invention. Section III of the opinion refers to those 25 as phantom quotes.
What sanction was imposed?
Counsel for the appellee must complete a continuing legal education course relating to AI or appellate practice within six months of the issuance of the opinion. No monetary sanction was imposed.
Why was there no fine?
Because the unreliable quotations were identified before oral argument, no monetary losses were claimed or established, opposing counsel never raised the issue with the court, and counsel took responsibility at oral argument and assured the court there was no intent to mislead. The court said it believed him.
If the court was that sympathetic, why sanction at all?
Because it considered itself compelled to. The opinion states that Virginia precedent calls for a sanction against one who files a brief violating Code section 8.01-271.1, and that although being the subject of the cautionary tale might be penalty enough, the court was compelled by statute and case law to impose a consequence.
Did the court find that AI was used?
It concluded that the source of the phantom quotes was some form of computer-assisted research that went awry, and framed its guidance around lawyers' use of AI. Counsel's assurance that the misquotes were unintentional was accepted.
What is Legal Ethics Opinion 1902?
A Virginia State Bar opinion, introduced for comment and noted in a footnote to this judgment, dealing with opposing counsel's responsibilities when confronted with a brief fraught with AI-hallucinated misstatements of law or fact. The opinion decides nothing about it.
Related briefings
- Florida court fines a lawyer $1,500 over AI citations
- Tenth Circuit sets a published AI citation standard in Lichfield v. Kubler
- Illinois court fines $1,500 per AI-hallucinated citation
- Pinsent Masons letter quoted an Insolvency Rule that does not exist
- North America AI regulation hub
- AI Regulation News hub
Sponsored Training
Practical AI training for regulated professionals, built around verification, documentation and a defensible process. See the courses.