Can a Lawyer Be Sanctioned for Using AI-Generated Fake Cases?
Recent decisions show that courts are no longer treating these errors as harmless technology mistakes. Depending on the circumstances, consequences can include monetary sanctions, disciplinary referrals, suspension from practice and other professional consequences. (New York Courts)
The important point is that using AI itself is not what triggers sanctions. The problem is submitting inaccurate material to a court without conducting the verification and professional review required by the applicable rules.
Can a lawyer legally use ChatGPT or another AI tool?
Generally, yes.
There is no blanket U.S. rule saying that lawyers are prohibited from using generative AI for legal work.
The American Bar Association has specifically addressed lawyers' use of generative AI and explains that existing professional-responsibility rules continue to apply when lawyers use these tools. The ABA's Formal Opinion 512 discusses competence and other ethical obligations arising from generative AI use. (American Bar Association)
The problem arises when a lawyer treats an AI-generated answer as if it were verified legal research.
An AI system may produce a citation that looks completely authentic but does not exist. It may also attach a genuine citation to a proposition the case never actually decided.
That distinction is critical.
A lawyer therefore needs to treat AI output as work product requiring verification—not as an authoritative legal source.
What is an AI hallucinated legal citation?
An AI hallucination occurs when a generative AI system produces information that sounds plausible but is false, unsupported or incorrectly presented.
In legal research, that can take several forms.
An AI system might:
invent an entirely nonexistent case;
provide the wrong case name;
give an incorrect reporter citation;
attribute a quotation to a real court that never said it;
combine language from different cases;
describe a real case as holding something it did not hold;
invent procedural history;
cite a real case from the wrong jurisdiction; or
present an outdated rule as current law.
The danger is obvious: a fabricated citation can look remarkably convincing.
That is why courts have repeatedly emphasized that attorneys must independently verify the authorities they submit.
What does Federal Rule 11 require?
For federal civil litigation, one of the most important rules is Federal Rule of Civil Procedure 11.
When an attorney presents a pleading, motion or other paper to a federal court, Rule 11(b) provides that the attorney certifies, after an inquiry reasonable under the circumstances, that the legal contentions are warranted by existing law or by a nonfrivolous argument for changing or extending the law. (Legal Information Institute)
That means the lawyer cannot simply argue:
“The AI generated the citation, so I am not responsible.”
The signature and filing belong to the attorney.
Rule 11 also allows courts to impose sanctions on an attorney, law firm or party responsible for a violation. Depending on the circumstances, sanctions can include nonmonetary directives, penalties paid into court and, in appropriate circumstances, payment of expenses and attorney fees. (Legal Information Institute)
Rule 11 therefore creates an important practical principle:
The technology used to prepare a filing does not eliminate the lawyer's responsibility for the filing.
What does the ABA say about technological competence?
The ABA Model Rules recognize that competence increasingly includes understanding relevant technology.
Model Rule 1.1 requires competent representation, while Comment 8 states that lawyers should keep abreast of changes in law and practice, including the benefits and risks associated with relevant technology. (American Bar Association)
The ABA's Formal Opinion 512 specifically addresses generative AI.
The opinion does not say that every lawyer must become an AI engineer. Instead, the lawyer needs a reasonable understanding of the capabilities and limitations of the particular AI technology being used. (American Bar Association)
That matters because knowing that generative AI can hallucinate legal authorities is itself part of using the technology responsibly.
What about the duty of candor toward the court?
Another major rule is ABA Model Rule 3.3.
Under Rule 3.3, a lawyer must not knowingly make a false statement of fact or law to a tribunal and must take reasonable remedial measures when necessary to correct a false statement previously made to the tribunal. (American Bar Association)
This becomes particularly important when an attorney discovers after filing that an AI-generated citation was wrong.
The problem does not necessarily end when the filing has been submitted.
If a lawyer discovers that a material legal statement or citation in a filing is false, the lawyer may have an obligation to correct the record depending on the applicable jurisdiction's rules and the circumstances.
The recent cases make this point especially clear.
Landberg v. City of New York: a major 2026 warning
One of the clearest 2026 examples is Landberg v. City of New York.
The New York Appellate Division, Second Department considered a brief prepared with the assistance of generative AI that contained nonexistent cases, fabricated quotations attributed to the New York Court of Appeals and misrepresentations about real cases. (New York Courts)
The court-imposed sanctions of $8,000 against the attorney and $2,500 against the law firm. The case was also highlighted by the ABA as a significant AI-related sanctions decision. (American Bar Association)
The court's reasoning is particularly important.
The attorney had access to ordinary legal research resources and professional obligations. The court therefore rejected the idea that AI use somehow excused the failure to verify the authorities.
This illustrates a broader rule:
AI assistance does not replace the lawyer's professional duty to know what the lawyer is filing.
Del Biaggio v. Bansen: verification cannot simply be delegated
A particularly important new case in 2026 is Del Biaggio v. Bansen, decided by the California Court of Appeal on July 10, 2026.
The attorney's appellate brief contained misstatements of case law, including fictional quotations. The court ultimately ordered attorney Carlton Floyd to pay $1,500 in sanctions. It also directed the clerk to forward the opinion to the State Bar and required the attorney to provide a copy of the opinion to his client and certify that he had done so. (California Courts)
The case is particularly valuable because it addresses the idea of human verification.
According to the court's opinion, the attorney had used generative AI and relied on a paralegal verification process. The court concluded that this approach did not satisfy the attorney's professional responsibility to review the authorities personally. (Reuters)
The lesson is powerful:
A lawyer cannot necessarily solve the AI-verification problem simply by telling a nonlawyer employee to check the citations.
The attorney remains responsible for the legal authorities used in the filing.
Malkeet Lnu v. Blanche: sanctions can become much more serious
Another 2026 case demonstrates how severe the consequences can become.
In Malkeet Lnu v. Blanche, the Ninth Circuit dealt with briefs containing nonexistent cases, misattributed quotations and serious misrepresentations of actual cases.
The attorneys had argued that the errors were innocent mistakes and had denied the possibility that AI was responsible.
The court imposed disciplinary consequences including six-month suspensions and $2,500 personal sanctions for each attorney, according to the ABA's case update. (American Bar Association)
This demonstrates why the issue is not simply about a small monetary penalty.
Depending on the facts and the jurisdiction, AI-related misconduct can become a professional-discipline issue.
Perez-Castillo v. Blanche
The Seventh Circuit's 2026 decision in Perez-Castillo v. Blanche provides another example.
The opening brief contained numerous AI hallucinations, including fabricated quotations, cases incorrectly identified as circuit precedent, erroneous legal propositions and factual assertions contradicted by the record. (American Bar Association)
This type of conduct creates multiple problems at once.
It can undermine:
the lawyer's credibility;
the client's position;
the court's ability to evaluate the case;
the integrity of the judicial process; and
the lawyer's professional standing.
The lesson is broader than simply “check your citations.”
Every substantive assertion produced or assisted by AI must be evaluated before it becomes part of a lawyer's filing.
Mata v. Avianca: the case that changed the conversation
The most famous example remains Mata v. Avianca, Inc.
In 2023, lawyers submitted a filing containing fabricated case authorities generated with the assistance of ChatGPT. The attorneys were sanctioned after the court determined that the cited cases did not exist and that the lawyers had failed to conduct the necessary verification.
The case became a landmark example of the danger of treating generative AI as a conventional legal research database.
It established a lesson that remains relevant in 2026:
A lawyer cannot rely on the apparent confidence or sophistication of an AI response as proof that the underlying authority exists.
The subsequent cases demonstrate that courts have not forgotten this lesson. (cdn.ymaws.com)
Park v. Kim: courts expect independent verification
In Park v. Kim, the Second Circuit dealt with an attorney who relied on ChatGPT research and cited nonexistent case law.
The attorney ultimately acknowledged that the cited authority had not been verified.
The matter was referred for further disciplinary consideration. (cdn.ymaws.com)
Again, the problem was not simply that an AI system made an error.
The central issue was the attorney's decision to submit the material without adequate verification.
Is a lawyer automatically sanctioned if AI makes a mistake?
No.
That distinction is extremely important.
A lawyer's use of AI does not automatically mean that a sanction is appropriate every time an AI-generated draft contains an error.
Courts examine the circumstances.
Questions may include:
Was the error material?
Was the citation nonexistent?
Was the authority misrepresented?
Did the attorney verify the citation?
Were there obvious warning signs?
Did the attorney conduct a reasonable inquiry?
Did the attorney knowingly make a false statement?
Did the lawyer discover the problem before or after filing?
If discovered after filing, did the lawyer promptly correct it?
Did the conduct burden the court?
Was the conduct repeated?
Was the attorney candid when questioned?
These factors can significantly affect the outcome.
What sanctions can a lawyer face?
Depending on the jurisdiction and circumstances, potential consequences may include:
Monetary sanctions.
A court may order an attorney or law firm to pay money.
Payment of opposing counsel's expenses.
Some rules permit recovery of expenses or attorney fees caused by improper conduct.
Sanctions payable to the court.
A court may impose a penalty intended to deter similar conduct.
Disciplinary referral.
A court may refer the attorney to a state bar or disciplinary authority.
Suspension.
Serious professional misconduct can potentially result in suspension from practice.
Disqualification or other litigation consequences.
Depending on the case and jurisdiction, courts may impose additional procedural consequences.
Damage to professional credibility.
Even when the monetary sanction is relatively modest, being identified in a published judicial opinion for submitting fabricated authorities can have significant professional consequences.
The 2026 decisions demonstrate that these consequences are not theoretical. (New York Courts)
Can a lawyer blame ChatGPT?
Usually, that is not a safe defense.
The important legal question is generally not:
“Who generated the sentence?”
It is:
“Who submitted the legal argument to the court, and did that person satisfy the applicable professional obligations?”
When an attorney signs and files a document, the attorney assumes professional responsibility for the contents under the applicable rules.
That is why statements such as “ChatGPT made the mistake” or “the AI generated the case” may explain how an error occurred, but they do not necessarily eliminate responsibility.
What should a lawyer do before citing an AI-generated case?
A simple verification protocol can dramatically reduce the risk.
First, search for the case independently using a reliable legal research source.
Second, confirm that the case actually exists.
Third, confirm the court.
Fourth, confirm the date.
Fifth, confirm the citation.
Sixth, open the actual opinion.
Seventh, locate the quoted language in the opinion.
Eighth, read enough surrounding context to determine whether the quotation means what the brief claims.
Ninth, verify that the case remains good law where that matters.
Finally, make sure the proposition attributed to the case is actually supported by the decision.
A citation should never be considered verified merely because an AI tool provides a citation number.
What if the lawyer discovers the AI error after filing?
This is one of the most important questions.
The lawyer should not simply ignore the problem.
The appropriate response depends on the jurisdiction, the type of filing, the significance of the error and the applicable ethical and procedural rules.
But Model Rule 3.3 illustrates the basic professional principle: a lawyer must not knowingly leave a false material statement before a tribunal and may have a duty to take reasonable remedial measures. (American Bar Association)
The Del Biaggio decision is especially instructive because the court considered the attorney's delayed correction of the erroneous material as an important part of the professional problem. (Reuters)
Therefore, discovering an AI-generated error should trigger an immediate review of:
what was filed;
how material the error is;
whether the court has relied on it;
whether opposing counsel has relied on it;
whether correction is required;
what the jurisdiction's professional rules require; and
whether the client must be informed.
Does every U.S. state follow the ABA Model Rules?
No.
This is another important limitation.
The ABA Model Rules are influential, but they are model rules, not automatically binding law in every state.
Individual states adopt their own professional-conduct rules, and those rules may differ.
For that reason, a lawyer facing a real disciplinary or sanctions issue should consult the rules applicable in the jurisdiction where the lawyer is practicing and where the filing was made.
The general principles discussed in this article are therefore not a substitute for jurisdiction-specific legal advice.
What does this mean for clients?
Clients should understand that AI use by a lawyer is not necessarily a red flag.
AI can be useful for administrative work, drafting assistance, document organization and other legitimate purposes.
The important question is whether the lawyer has appropriate systems for reviewing AI output.
A client can reasonably ask:
“Do you use AI in my case?”
“If so, for what purpose?”
“How are AI-generated legal authorities verified?”
“Who reviews the final filing?”
“Are confidential client materials being entered into an AI system?”
These questions do not accuse a lawyer of misconduct. They simply help the client understand how technology is being used in the representation.
A practical AI verification checklist for lawyers
Before filing any AI-assisted document, an attorney should consider this checklist:
Case existence: Does every cited case actually exist?
Citation accuracy: Is the case name, court, year and reporter citation correct?
Quotation accuracy: Does the quoted language appear word-for-word in the actual opinion?
Holding accuracy: Does the case actually establish the proposition being attributed to it?
Jurisdiction: Is the authority from the correct jurisdiction?
Current status: Has the authority been overruled, superseded or otherwise limited?
Factual accuracy: Are factual statements supported by the record?
Procedural history: Has AI invented or distorted what happened earlier in the case?
Primary-source verification: Has the lawyer reviewed the actual opinion or authoritative source?
Human review: Has a qualified attorney reviewed the final material?
Candor: If an error was discovered, has the lawyer taken whatever corrective action the applicable rules require?
This is not an exhaustive ethics checklist, but it is a useful starting point.
The bigger legal lesson
The emerging U.S. case law does not suggest that courts are banning AI from legal practice.
The trend is more nuanced.
Courts are essentially saying:
Use technology if it helps you—but remain responsible for what you submit.
That distinction is likely to become increasingly important as AI becomes integrated into legal research platforms, drafting software and law-firm workflows.
The recent cases also show that the danger is not limited to completely fabricated cases.
A lawyer can create a serious problem by:
citing a real case for the wrong proposition;
using a fabricated quotation from a real case;
misstating a holding;
relying on outdated law;
submitting unsupported facts; or
failing to correct a known error.
In other words, “the citation exists” is not the same as “the citation supports my argument.”
Frequently Asked Questions
Is it illegal for lawyers to use ChatGPT?
Not automatically. Lawyers may use AI tools, but they remain subject to applicable professional-conduct rules, confidentiality obligations, competence requirements and court rules. (American Bar Association)
Can ChatGPT be used for legal research?
It can be used as an assistive tool, but AI-generated authorities should be independently verified against reliable primary legal sources before being relied upon in legal work.
Can a lawyer be sanctioned if ChatGPT invents a case?
Potentially, yes, particularly if the lawyer submits the nonexistent authority without reasonable verification or otherwise violates court or professional rules. Recent 2026 decisions illustrate this risk. (New York Courts)
Can a lawyer say, “AI made the mistake”?
That explanation may describe how the error happened, but it does not automatically eliminate the lawyer's responsibility for the filing.
What is the most important AI rule for lawyers?
There is no single nationwide “AI lawyer rule.” Existing rules concerning competence, candor, reasonable inquiry, confidentiality and supervision can apply to AI-assisted legal work. ABA Model Rule 1.1 and Rule 3.3 are particularly relevant. (American Bar Association)
What should a lawyer do if an AI-generated citation is discovered after filing?
The lawyer should promptly assess the error and applicable procedural and professional rules and take any required corrective action. The duty of candor can continue after the original statement has been made. (American Bar Association)
Conclusion
Can a lawyer be sanctioned for using AI-generated fake cases? Yes—but the critical issue is not simply the use of AI.
The real legal problem arises when an attorney submits fabricated citations, false quotations, materially inaccurate legal authorities, or unsupported legal arguments without conducting the reasonable investigation, professional review, and verification required by applicable court and professional-conduct rules.
These 2026 decisions demonstrate that courts are imposing real consequences when attorneys submit AI-generated legal errors without adequate verification. (New York Courts)
Depending on the circumstances, those consequences can include monetary sanctions, disciplinary referrals, suspension, and other professional consequences.
Official / Authoritative Sources
Federal Rule of Civil Procedure 11 — Cornell Legal Information Institute — federal requirements concerning reasonable inquiry and sanctions. (Legal Information Institute)
ABA Model Rule 3.3 — Candor Toward the Tribunal (American Bar Association)
ABA Formal Opinion 512 — Generative Artificial Intelligence Tools (American Bar Association)
New York Courts — Landberg v. City of New York (New York Courts)
California Courts — Del Biaggio v. Bansen (California Courts)
Legal Disclaimer
This article is for general educational and informational purposes only. It is not legal advice and does not create an attorney-client relationship. U.S. professional-conduct rules, court rules and sanctions standards can vary by jurisdiction and by the facts of a case. Anyone facing a real legal, disciplinary or court-sanctions issue should consult a qualified attorney licensed in the relevant jurisdiction.
Comments
Post a Comment
Comment should be in the form of message and helpful to the viewers