Hrdlichka v. Bengston: Arkansas Court of Appeals Dismisses Appeal Over AI-Fabricated Citations in Historic First | Advocate Prakhar

Case at a Glance

Background

Samantha Bengtson filed a civil complaint in Benton County Circuit Court alleging that Patrick Hrdlichka had committed a second-degree battery against her in February 2022 and sought civil damages under Arkansas Code Annotated section 16-118-107 (civil action by crime victims). After a bench trial, the circuit court awarded Bengtson $10,000 in compensatory damages plus interest and costs. Hrdlichka, proceeding pro se, appealed.

Hrdlichka’s appellate brief raised five substantive points: incorrect application of the statute, lack of felony evidence, insufficient compensatory damages proof, inconsistency with established Arkansas jurisprudence, and constitutional equal protection arguments. The brief also contained numerous procedural deficiencies — the table of contents and points on appeal were not arranged correctly, the statement of facts lacked record citations for most of the narrative, and the argument section lacked proper subheadings. However, what caught the court’s attention most directly was that multiple cases cited in the brief “do not exist” and “are not found in the Arkansas Reports, the South Western Reporter, or any recognised legal database.”

The AI Issue

The court acknowledged explicitly that this was “our first opportunity to consider the impact of fictitious cases being submitted to this court, an issue that has gained national attention in the rising availability of artificial intelligence.” It quoted directly from academic commentary: “Citing nonexistent case law or misrepresenting the holdings of a case is making a false statement to a court; it does not matter if generative AI told you so.” The court then issued a comprehensive warning applicable going forward: AI use without verification risks dismissal; fictitious citations violate Rule 4-2; and the duty of verification is not reduced by the fact that the filer is pro se.

What the Court Decided

• Appeal dismissed for failure to comply with Arkansas Supreme Court Rule 4-2 due to fabricated citations (citations do not conform to required format and do not correspond to any real authority) and multiple additional briefing deficiencies.
• Court stated it “cannot evaluate arguments predicated on nonexistent precedent” — fabricated citations deprive the court of any meaningful ability to review the legal arguments.
• Rule 11 of the Arkansas Rules of Appellate Procedure (equivalent to federal Rule 11) applies: by signing and filing, an attorney certifies the submission is “warranted by existing law.”
• Future filings with AI-hallucinated citations may be struck; sanctions including dismissal may be imposed.
• Court noted that using AI is not itself prohibited, but using AI without verification — so that fabricated cases are submitted as real precedent — violates the duty of candour.

“Citing nonexistent case law or misrepresenting the holdings of a case is making a false statement to a court. It does not matter if [generative AI] told you so.”

— Arkansas Court of Appeals, Hrdlichka v. Bengston (2026 Ark. App. 205), quoting academic commentary

The India Angle

Indian Law Equivalent

In India, the equivalent of a dismissal for briefing deficiencies containing fabricated citations would occur under Order XLII Rule 1 CPC read with the High Court Rules, which give appellate courts the power to dismiss non-compliant appeals. The Supreme Court and High Courts have inherent jurisdiction to dismiss appeals or petitions filed with fabricated authorities under their general case management powers. The protection of the court’s dignity and the integrity of the legal process are both engaged when fabricated case citations are submitted — these engage not just procedural rules but the courts’ contempt jurisdiction under the Contempt of Courts Act, 1971.

Bar Council Rules

The Arkansas ruling’s core statement — “it does not matter if AI told you so” — is the clearest possible formulation of the professional responsibility principle applicable in India and everywhere else. BCI Rule 49 makes it the advocate’s personal responsibility to ensure that every statement of law made to a court is accurate. Delegating that responsibility to an AI tool that the advocate does not verify is not a defence — it is the violation itself.

Practical Advice for Indian Advocates

• Even a first-time AI-hallucination violation in an appellate court can result in dismissal of the appeal — there is no automatic “first warning” entitlement for pro se litigants or advocates appearing in Indian High Courts or the Supreme Court.
• Multiple simultaneous briefing deficiencies (formatting errors, missing record citations, AI-fabricated cases) compound each other — a brief with all these problems signals to the court that the entire document was generated by AI and filed without any human review.
• Pro se litigants in India who intend to use AI to draft appeals should be counselled that dismissal for AI-related briefing failures is a growing risk in international jurisdictions, and that the same risk is likely to materialise in India as AI use spreads in pro se litigation.

Quick Takeaways

• Appellate courts can and do dismiss entire appeals for AI-fabricated citations — no prior warning or lesser sanction is legally required before dismissal is appropriate.
• Arkansas joins a growing roster of state and federal courts that have issued first-impression rulings on AI hallucinations, building a national jurisprudential consensus that the duty of verification is absolute.
• Multiple simultaneous briefing deficiencies characteristic of unreviewed AI output (formatting errors + citation errors + missing record references) collectively establish that a brief was filed without meaningful human review.

Deep Dive: State-by-State AI Jurisprudence — Why Every Jurisdiction Is Writing Its Own First Chapter

The Hrdlichka v. Bengston ruling is significant for its position in the emerging national jurisprudence on AI hallucinations. The court expressly acknowledged that this was its first encounter with AI-fabricated citations and treated the occasion as an opportunity to establish a clear standard for future cases. This pattern — a court issuing a comprehensive warning when it first confronts the problem — is now visible in courts across multiple US states, in Canada (Galang v. Canada, BC Supreme Court; H.B. v. Conseil des Écoles, Ontario HRTO), in the UK (Tuyunuklu v. Akmal, UKUT), and in Florida’s string of decisions.

The Arkansas court’s decision to quote academic commentary — the Grossman, Grimm & Brown Judicature article — alongside its own analysis reflects the cross-jurisdictional convergence of AI-hallucination standards. Courts are drawing on the same academic sources, citing each other across state and national lines, and building toward a common understanding: AI-generated text is unreliable for legal citations; the advocate who submits it without verification is professionally responsible for the error; no explanation — time pressure, lack of resources, good intentions — constitutes a valid defence.

For Indian advocates and courts, the value of tracking this emerging international jurisprudence is that it provides a framework for what Indian courts are likely to adopt when they encounter the same problem. India’s legal system has consistently looked to Anglo-American jurisprudence for analogies in novel situations. When an Indian court first confronts an AI-hallucination case — likely in a High Court within the next one to three years — it will have access to rulings from Arkansas, Florida, New York, Canada, and the UK, all converging on the same standard. The advocate who understands that standard in advance will be best positioned to advise clients and manage professional risk.