Detecting AI Misconduct by Opposing Counsel Is a Lawyer’s Duty

Much discussion about artificial intelligence has centered on a lawyer’s duty to competently and ethically use it. A recent California court decision raises a critical, additional question: Do attorneys have a responsibility to detect and report an opponent’s use of AI, especially when that use results in fabricated or “hallucinated” legal authority?

In Noland v. Land of the Free, L.P., the California Court of Appeal, Second District, determined that proactive detection is best practice and a form of financial self-defense. The court’s denial of fees establishes that attorneys who fail to identify AI fraud may not qualify for recovery, even when opposing counsel’s misconduct deserves sanctions.

As Noland suggests, if you miss it, you may lose your opportunity to recover for it.

‘Enhancing’ a Brief

In Noland, the plaintiff’s counsel used generative AI to “enhance” his appellate brief. The outcome was disastrous: 21 of the 23 case quotations in his opening brief were fabrications. The appellate court discovered this, then initiated an order to show cause and imposed a $10,000 sanction on the plaintiff’s attorney as well as a referral to the State Bar of California.

However, when opposing counsel sought attorneys’ fees as a sanction, the court denied the request. The reason was explicit: Opposing counsel “did not alert the court to the fabricated citations and appeared to have become aware of the issue only when the court issued its order to show cause.”

This refusal to award fees sends a clear message that failing to detect an opponent’s AI-caused ethical lapse could leave one without a means to recover the time and costs incurred in debunking the false claims.

California’s Ethical Framework

Although the California Rule of Professional Conduct doesn’t explicitly mandate checking an opponent’s AI citations, existing rules establish a clear expectation of diligence.

Rule 8.3 requires that a lawyer inform the state bar or a tribunal immediately when they know of credible evidence that another lawyer has engaged in conduct involving dishonesty, fraud, deceit, or reckless or intentional misrepresentation. Under Rule 8.4(c), submitting a brief with hallucinated citations falls within the definition of misconduct involving misrepresentation and deceit.

The Noland court implied that if opposing counsel had exercised due diligence and detected the fake citations, they would have had a clear duty under Rule 8.3 to report the misconduct. Failing to detect the issue was interpreted as a lack of diligence, leading directly to fee denial.

Noland indicates that courts now view the identification of AI-generated fraud as a crucial element of professional competence.

The court’s fee denial suggests it viewed this situation as a serious failure to uphold the adversarial process’s integrity. The court expressed frustration at having to expend “excessive time on this otherwise straightforward appeal to attempt to track down fabricated legal authority.”

Overlooking fraudulent arguments leads to forfeiting the chance to recover attorneys’ fees and wastes client resources on debunking specious claims, while also risking judicial criticism for imposing unnecessary work on the court.

Safeguarding Against Misconduct

In light of Noland, California attorneys must adopt a strategic, skeptical approach to all opposing filings that take steps to:

Establish a zero-tolerance policy for unverified citations. You can’t assume an opponent’s citations are accurate. Treat every citation as a potential AI hallucination until thoroughly verified.

Your team must manually verify every case, statute, and quotation cited. Confirm the source’s existence and validity using reputable legal research services (Bloomberg Law, Westlaw, LexisNexis, etc.) and verify that the quoted language appears precisely in the cited authority and supports the claim.

Watch for AI hallucination red flags. Train your team to identify typical indicators of misleading citations:

• Plausible but nonexistent cases. A citation may look legitimate, for example, Smith v. Jones (2024) 10 Cal.App.5th 100—but yield no results.
• Misleading quotes. A quote may exist, but it may be from a different case or cited for a proposition that the source refutes.
• “Off” language. Generative AI often produces overly formal, repetitive, or slightly unusual language compared with typical legal writing.

Promptly report and seek sanctions. Bringing the issue to the court’s attention is essential to your ethical duty and fee recovery. If you discover a pattern of fake citations, Rule 8.3 requires reporting to the tribunal or the state bar without undue delay.

Documentation is also critical. Maintain a clear, verifiable record of your research showing that the citations are false and detailing the time and resources wasted.

File a motion for sanctions and costs. Alert the court immediately to the fabricated authority, explain its impact on the court’s time and your client’s resources, and request appropriate sanctions—including reimbursement for fees spent debunking false research.

Lawyers must adopt a higher level of skepticism and diligence in response to growing dependence on AI. Recognizing proactive detection as a critical risk management and litigation recovery strategy is essential, as mandated by the Rules of Professional Conduct and supported by developing case law.

Proactive detection goes beyond a mere courtesy to the court; it is fundamental to guaranteeing accountability and safeguarding client interests from an opponent’s potential ethical or technological failures.

The case is Noland v. Land of the Free, L.P., Cal. Ct. App., 2d Dist., No. B331918, decided 9/12/25.

This article does not necessarily reflect the opinion of Bloomberg Industry Group, Inc., the publisher of Bloomberg Law, Bloomberg Tax, and Bloomberg Government, or its owners.

Author Information

Joe Stephens is a consulting attorney and legal AI expert at Steno.

Write for Us: Author Guidelines