News

The Sixth Circuit’s recent decision in Whiting v. City of Athens, No. 24-5918/5919, 2026 U.S. App. LEXIS 7479 (6th Cir. Mar. 13, 2026) is another cautionary tale about generative AI in litigation, but goes beyond the now-familiar “fake case citation” problem. The opinion highlights several less-discussed issues: what courts may require lawyers to disclose about their use of AI, how hallucinations can affect factual statements about a case (not just legal authorities), and how courts can rely on their inherent authority—not just Rule 38—to sanction misconduct that threatens the integrity of judicial proceedings.

What Happened

The appeal arose from litigation related to a dispute surrounding a municipal fireworks event. But during its review of the appellate briefs, the Sixth Circuit discovered serious problems. The filings cited numerous nonexistent cases, quoted authorities for propositions they did not support, and misrepresented the record and procedural history. The Whiting court ultimately identified more than two dozen fake citations and factual misstatements leading to issuance of a show-cause order requiring counsel to produce copies of every cited authority and to disclose detailed information about how the briefs were prepared. But rather than acknowledging the errors, the lawyers challenged the legitimacy of the court’s show-cause order and did not meaningfully address the fabricated citations or misrepresentations.

Case Law and Procedural Hallucinations Render Appeals “Frivolous” and Sanctionable

The Sixth Circuit imposed sanctions under Federal Rule of Appellate Procedure 38, concluding that the appeal was “frivolous as argued.” The court drew an important distinction between two types of frivolous appeals. An appeal is “frivolous as filed” when the lower court is so plainly correct that no reasonable argument could support reversal. An appeal is “frivolous as argued” when arguable issues may exist, but the appellant engages in misconduct in arguing the record. For example, when lawyers cite nonexistent authorities, mischaracterize case law, or misrepresent the record, which is precisely what happened here.

The Whiting court clarified that despite the missing language in Rule 38, it indeed recognizes frivolous-as-argued appeals because such arguments could never support reversal. It then emphasized that inventing case law is itself a misrepresentation of law, making the appeal frivolous. The court also noted that the same conduct could potentially support a finding that the appeal was frivolous as filed, because hallucinated cases often appear when lawyers attempt to support legal propositions for which no authority exists.

In addition to Rule 38 sanctions, the court awarded penalties pursuant to its own plenary power to sanction a frivolous litigant. Unlike Rule 38, a court’s inherent powers allow it to sanction bad-faith conduct that undermines the judicial process, including misleading the court or abusing the adversarial system. Courts have this authority because they must be able to protect the legitimacy and reliability of their proceedings.

The Sixth Circuit found that the appellant’s lawyers engaged in bad faith by relying on false case law in using the court system to force a result they could not otherwise obtain under applicable law. The court pointed out that when the lawyers were given an opportunity to explain how the fake citations appeared in the briefing, they failed to provide an explanation and instead challenged the court’s show cause order as invalid. Explaining that Rule 38 was not “up to the task” of sanctioning such egregious conduct, the Sixth Circuit employed its inherent authority to issue the following sanctions to appellant’s counsel:

• Reimburse appellees in full for their reasonable attorneys’ fees on appeal,
• Jointly and severally pay double costs to appellees for costs incurred under 28 U.S.C. § 1920,
• Separately and individually pay $15,000 to the court’s registry, and
• The court’s order would be forwarded to the chief judge to consider disciplinary proceedings.

The court also made clear that the attorneys’ conduct could have warranted more severe consequences, including potential disbarment from practice before the court or contempt sanctions.

What This Case Adds to the AI-in-Law Conversation

AI Hallucinations Are Not Just About Fake Cases. Much commentary on AI errors focuses on fabricated case citations. But Whiting highlights another risk: hallucinations about the case itself. Generative AI can fabricate or distort procedural history, descriptions of the record, and even the factual background on the lawyer’s own case. The Whiting court treated misrepresenting the record or procedural posture as seriously as citing nonexistent authorities. Lawyers using AI in drafting must not only verify the legal authorities cited, but also any summarization of the record created with assistance from AI.

Ghostwriting Rules Apply Beyond the Pro Se Context. Whiting also reinforces a developing theme in AI case law: courts are increasingly evaluating AI-assisted drafting through the lens of ghostwriting. Federal courts have generally condemned ghostwriting by an attorney for a pro se plaintiff as unethical and evasive of the responsibilities imposed on attorneys. Last year, the court in Buckner v. Hilton Glob., No. 3:24-cv-375-RGJ, 2025 U.S. Dist. LEXIS 117533 (W.D. Ky. 2025 June 20, 2025) explained that AI use in drafting can resemble ghostwriting when it functions as an undisclosed substitute drafter. The Sixth Circuit now clarifies that ghostwriting’s problems of concealed authorship and diminished accountability are not limited to the pro se context. If the purported author of a filing can avoid responsibility because another person (or tool) actually drafted or generated the filing, the court’s ability to enforce candor and professional obligations is undermined. The Whiting court suggests that the legal issue with AI-assisted drafting is not whether AI is literally “writing” the brief, but whether undisclosed AI use risks the same problems as ghostwriting. Before filing with the court, lawyers using AI should always check local rules and court procedures to determine if disclosure of such use is necessary and if so, in what form.

The Scope of Information a Court May Require Counsel to Disclose. When briefing appears unreliable, courts may ask litigants to explain the following, without concern for waiver of attorney client or work product privilege:

• whether generative AI was used,
• who drafted the briefs,
• whether ghostwriting occurred,
• which portions relied on AI, and
• how the authorities were verified.

While AI is no substitute for reading and substantiating the authority relied upon, it can be leveraged to streamline a double-check process. Lawyers should prompt their AI tools to pull all of the cited authority into a folder for review, evaluating it against the propositions for which it was cited, and make suggestions regarding whether conflicting authority exists.

The Right Response When Errors Occur. The Whiting opinion also underscores a professional responsibility point: if lawyers discover they have cited a nonexistent case, the appropriate response is to acknowledge the mistake, apologize, and correct the record. Attempting to deflect responsibility or challenge the court’s authority can significantly worsen the situation.

Takeaways for Lawyers Using AI

Whiting highlights the particular risks of AI misuse in an appellate context. Unlike trial courts, appellate courts rely almost entirely on the parties’ briefing to understand both the governing law and the underlying record. When that briefing contains fake authorities or distorted case descriptions, the court’s ability to adjudicate the appeal is directly compromised. The Sixth Circuit’s decision makes clear that courts of appeals will not hesitate to look behind the filing itself, inquire into how it was prepared, and may increasingly rely on their inherent authority to sanction conduct that undermines the reliability of appellate advocacy. In that sense, Whiting is less about technology than about accountability and ensuring lawyers do not use that technology to subvert the long-standing duties of candor, accuracy and responsibility.