Blackburn v. United States (Oklahoma 2026): Court Denies Motion in Limine After Finding Mischaracterized AI Citations — Warns of Rule 11 Consequences | Advocate Prakhar

Case at a Glance

Background

Plaintiff Tyeashia Blackburn filed a motion in limine in an FTCA medical malpractice case against the Oklahoma City VA Health Center seeking to enforce requirements on the testimony of defendant’s non-retained treating physicians. To support her motion, Blackburn cited three authorities. The court examined each in detail.

First, Davoll v. Webb, 194 F.3d 1116 (10th Cir. 1999), was cited correctly, but the pinpointed page (1138) discussed treating physician testimony as a lay witness, not as a non-retained expert — so the citation did not support the proposition for which it was cited. Second, Muscogee (Creek) Nation v. Oklahoma was cited for the proposition that “a treating physician may not testify to opinions acquired or developed in anticipation of litigation” — but that case nowhere discussed treating physicians or their testimony. Third, a citation to “Hall v. United States, 2018 WL 1620923 (D. Colo. Apr. 3, 2018)” was provided, but Westlaw’s identifier for that citation directed the court to Poppino v. Berryhill, a Social Security disability appeal from New Mexico with no relevance to VA treating physicians.

The court denied the motion both because it lacked the specificity required to establish entitlement to relief and because none of the cited authorities actually supported the argument. The court then addressed the citation accuracy issue as a separate matter.

The AI Issue

The court noted that the three citation errors — correct citation with wrong page, real case with no relevance, and Westlaw identifier leading to a completely different case — constitute “mischaracterizations of case law, whether or not the product of generative artificial intelligence,” and implicate Rule 11(b). The court cited the Tenth Circuit’s recent guidance in Dodds v. Bridges (2026) that pro se litigants have “the responsibility to ensure that citations to legal authority are not fabrications but instead point to real cases that at least arguably stand for the propositions for which they are cited.”

What the Court Decided

• Mischaracterizing case law — whether by citing the wrong page of a real case, citing a real case for an inapplicable proposition, or providing a Westlaw citation that redirects to an unrelated case — implicates Rule 11(b) obligations regardless of whether AI was used.
• Rule 11(b) requires that legal contentions be warranted by existing law — a case that exists but says nothing about the cited proposition does not satisfy this standard.
• The Tenth Circuit has clarified that the verification responsibility applies to pro se litigants: citations must point to real cases that “at least arguably stand for the propositions for which they are cited.”
• The motion was denied on both the substantive (non-specific) and evidentiary (mischaracterized citations) grounds.

“Such mischaracterizations of case law, whether or not the product of generative artificial intelligence, implicate consideration of Plaintiff’s representations for purposes of Federal Rule of Civil Procedure 11(b) and (c).”

— Western District of Oklahoma, Blackburn v. United States, April 7, 2026

The India Angle

Indian Law Equivalent

In Indian practice, citing a case for a proposition it does not stand for is equally a misrepresentation to the court, even if the case exists. The Supreme Court in Union of India v. Ibrahim Uddin (2012) 8 SCC 148 emphasized that advocates must cite cases accurately for the propositions they actually stand for, and not selectively quote from judgments to create a misleading impression of the holding. AI-generated citation mischaracterization — presenting a real case as supporting a proposition it does not — falls squarely within this prohibition.

Bar Council Rules

BCI Rules, Chapter II, Rule 9 prohibits advocates from doing anything that tends to mislead the court, which would encompass citing cases for propositions they do not support. Rule 22’s requirement of factual accuracy in representations extends to accurate description of case holdings — a case “citation” that misrepresents what the case says is a factual misrepresentation about the law.

Practical Advice for Indian Advocates

• When using AI to identify supporting cases, always read the full relevant passage of the cited judgment — AI tools often assign the right case name to the wrong proposition, which is just as misleading as citing a non-existent case.
• Be especially careful with Westlaw and similar database cross-referencing: AI tools sometimes generate accurate reporter citations that correspond to entirely different cases than the one the tool describes.
• Before filing any brief, run each citation through the following check: Does the case exist? Does it say what I claim it says? Is the page I cite the relevant passage? This three-step check eliminates the most common AI mischaracterization errors.

Quick Takeaways

• Real cases cited for wrong propositions are just as sanctionable as completely fabricated cases.
• Rule 11(b) liability does not require AI use — mischaracterization from any source triggers it.
• A Westlaw citation that redirects to the wrong case can generate serious professional liability.

Deep Dive: The Mischaracterization Variant of AI Hallucination

The Blackburn case illuminates the mischaracterization variant of AI hallucination — one that is in some ways more insidious than complete fabrication. When an AI tool invents a case entirely, sophisticated practitioners often catch the error because the case simply cannot be found. But when an AI tool correctly names a real case, provides an accurate reporter citation, and then describes the case as standing for a proposition it does not support, the error survives basic citation-existence checks. A practitioner who confirms that Muscogee (Creek) Nation v. Oklahoma exists and is a published Tenth Circuit case may not then read the actual opinion to verify that it addresses treating physician testimony — and so the mischaracterization reaches the court undetected.

The Westlaw-identifier-to-wrong-case error is a particularly revealing AI artifact. When AI tools generate citations, they sometimes correctly identify the volume and page number of a reporter but attribute that citation to the wrong case, because the AI’s training data associated those coordinates with a different decision. When the court ran the Westlaw identifier for “Hall v. United States, 2018 WL 1620923,” it found Poppino v. Berryhill — a Social Security case from New Mexico that has nothing to do with VA treating physicians. This is a classic hallucination pattern: real coordinates, wrong case.

The court’s formulation — “whether or not the product of generative artificial intelligence” — reflects an important judicial attitude. Courts are not going to spend time determining whether a citation error originated from AI or from human carelessness; what matters is whether the citation misrepresents the state of the law. This approach is correct and efficient: the harm to the adversarial process is the same regardless of the cause, and Rule 11’s objective standard does not require proof of AI use.

For Indian practitioners, the Blackburn framework is particularly useful as a practice management tool. Rather than asking “did I use AI for this citation?” advocates should ask “does this citation actually support this proposition?” for every case in every brief, regardless of how it was researched. This proposition-by-proposition verification test, applied universally, eliminates the hallucination risk without requiring separate workflows for AI-generated and non-AI research.