Curry v. Capital One Finance (NC 2026): Pro Se FCRA Case Dismissed for AI-Generated Fictitious Statutory Provisions | Advocate Prakhar

⚡ Case Digest

Curry v. Capital One Auto Finance — E.D. North Carolina, March 26, 2026

A pro se FCRA plaintiff’s first proposed amended complaint relied on fictitious provisions of the North Carolina Identity Theft Protection Act that do not exist. Despite being warned in a prior order about AI-generated fabrications and the court’s intolerance of “computer generated morass,” her second proposed amended complaint also contained fictitious citations. The court dismissed the case and denied leave to amend.

Why it matters: AI hallucinations extend beyond case citations to fabricating statutory provisions — a court will not permit amendment if the proposed new claims are built on non-existent law.

Category: AI Hallucination & Sanctions  |  Jurisdiction: USA (North Carolina)  |  Read time: 6 min

Case at a Glance

Full CitationCurry v. Capital One Auto Finance, No. 5:25-CV-164-BO-KS (E.D.N.C.), March 26, 2026
CourtUnited States District Court, Eastern District of North Carolina, Western Division
DateMarch 26, 2026
AI Tool / IssueAI-generated first amendment cited fictitious NCITPA provisions; AI-generated second amendment contained fictitious case citations; amendment denied after prior warning
OutcomeOriginal complaint dismissed on the merits; leave to file second amended complaint denied as futile due to fictitious citations; case ended

Background

Shanierra Curry filed a complaint in Wake County Small Claims Court alleging violations of the Fair Credit Reporting Act. Capital One removed the action to federal court. In July 2025, the court denied Curry’s first motion to amend, finding the proposed complaint sought to assert new claims based on fictitious provisions of the North Carolina Identity Theft Protection Act — statutory text that does not exist. The court stated it would not tolerate “computer generated morass that only serves to waste court time and resources” and warned that it had concluded Curry likely used AI to draft the first proposed amended complaint.

In March 2026, Curry renewed her motion to amend, attaching a second proposed amended complaint. The court examined the second proposed complaint and found it also contained fictitious citations. Since Curry had already been warned once and the second proposed amendment still contained fabricated material, the court denied the second amendment motion as futile and simultaneously dismissed the original complaint for failure to state a claim, as Curry had not shown that Capital One was a furnisher of credit information rather than merely a user of credit reports.

The AI Issue

This case uniquely illustrates AI hallucination at the statutory level, not just the case citation level. AI tools do not only fabricate case names — they can fabricate entirely non-existent statutory provisions, citing them with correct-looking section numbers within real statutes. A party who builds their claim on non-existent statutory provisions has no viable cause of action, and no amendment that cites the same non-existent provisions can cure the deficiency. The court’s futility finding applied equally to the second proposed amendment because it too contained fabricated authority, making the proposed new claims legally baseless.

What the Court Decided

  • A proposed amended complaint that relies on fictitious provisions of a real statute is futile and will be denied leave to amend — the claims built on non-existent law have no legal basis regardless of their factual merits.
  • Once a court has warned a party about AI-generated fabrications and the party files a second proposed amendment also containing fabricated citations, the futility finding and denial of amendment are appropriate.
  • The original FCRA claims failed because Curry did not plausibly allege that Capital One was a furnisher (rather than user) of credit information, and the court dismissed the complaint on this independent ground.
  • The combination of meritless original claims and repeated AI-generated fabrications in amendment proposals justified terminating the case through dismissal rather than allowing further amendment attempts.

“The Court will not tolerate ‘computer generated morass that only serves to waste court time and resources.'”

— Quoting Thornock v. Corp. of the Presiding Bishop, adopted by E.D. North Carolina in Curry v. Capital One, March 26, 2026

The India Angle

Indian Law Equivalent

Indian courts have the power to reject amendments to plaints under Order VI Rule 17 CPC where the proposed amendment would not disclose a cause of action or would be futile. If a proposed amendment cites non-existent statutory provisions — as in Curry — the court would reject the amendment under Order VI Rule 17 as not bona fide or as futile. More broadly, filing a plaint or amendment that cites fabricated statutory provisions could constitute a fraud on the court, empowering the court to dismiss the suit under its inherent jurisdiction under Section 151 CPC.

Bar Council Rules

An advocate who presents a plaint or amendment relying on statutory provisions that do not exist violates BCI Rule 22’s requirement of factual accuracy and Rule 9’s prohibition on misleading the court. In the context of statutory fabrication, the advocate is not only misrepresenting case law but misrepresenting the substantive law itself — a graver breach that would warrant more serious disciplinary action.

Practical Advice for Indian Advocates

  • When using AI to identify applicable statutory provisions, always verify the provision against the official text of the statute as published by the relevant authority — AI tools frequently fabricate plausible-sounding section numbers that do not exist in the actual legislation.
  • Be particularly careful with AI-generated references to state legislation, regulations, and rules — these are less well-represented in AI training data than central statutes, making hallucinated provisions more common.
  • If a client asks you to rely on a statutory provision you are not familiar with, look it up directly in the official statutory text before accepting the client’s representation — fabricated provisions from AI tools are now being presented to advocates as real law by clients who generated them using AI.

Quick Takeaways

  • AI hallucinations can target statutory provisions, not just case citations — both are equally fatal to a legal claim.
  • A proposed amendment built on non-existent law is futile and will be denied regardless of factual merit.
  • Repeated AI fabrications after an explicit court warning eliminate any possibility of further amendment.

Deep Dive: Statutory Hallucination — A Distinct and More Dangerous AI Error

Curry v. Capital One highlights a variant of AI hallucination that deserves separate attention: the fabrication of statutory provisions. When an AI tool cites a non-existent case, a diligent practitioner can usually catch the error by running the citation through a legal database. But when an AI tool cites a plausible-sounding section number within a real statute — such as a provision of the North Carolina Identity Theft Protection Act that does not actually exist — the error is much harder to detect without reading the full statutory text.

Statutory hallucinations are particularly dangerous because they can drive the entire cause of action. In Curry, the proposed amendment did not merely cite a fake case for a proposition — it claimed a whole new set of claims under a statutory framework that did not exist. Even if all the other elements of the proposed amendment were perfectly valid, claims premised on non-existent statutory provisions have no legal foundation and will be dismissed as futile regardless of the underlying facts.

The prevalence of statutory hallucinations is likely to grow as AI legal tools are increasingly used for research in specialized areas of state law, consumer law, environmental law, and regulatory compliance — areas where the statutory landscape is complex, frequently amended, and variable across jurisdictions. AI models trained predominantly on case law and federal statutes may have less reliable knowledge of state statutory schemes, making hallucinated provisions more probable in exactly the areas where practitioners are most likely to rely on AI research assistance.

For Indian advocates, statutory hallucination is a significant emerging risk. India’s legislative landscape includes hundreds of central and state acts, with multiple amendments, gazette notifications, and subordinate legislation. AI tools may produce plausible-sounding references to sections, sub-sections, and rules that do not exist or have been repealed. The safest approach is always to verify statutory provisions against the official text available on India Code (indiacode.nic.in), the Ministry of Law portal, or the relevant official gazette — not to rely on AI output for the existence or wording of any legal provision.

What Our Clients Say

Chat on WhatsApp Call Now
Scroll to Top