Aziz v. USA: AI-Enabled Filing Avalanche + Fabricated Citations — Court Imposes 4-Page Filing Limit

Background: Related Litigation and a Filing Avalanche

Ali Behroz Aziz and Shinkay Aziz filed suit in the District of Maryland against the United States, the Maryland Attorney General’s Office, Bezhan Aziz (a family-relation defendant), and Jonathan Kagan. The case is related to a prior action — Aziz v. Aziz, No. 22-cv-2834 — suggesting a pattern of ongoing civil litigation involving the same parties.

Within three weeks after the court issued a prior opinion on several of Plaintiffs’ motions and an order to show cause, Plaintiffs filed seven additional motions, including motions to convert, amend, dismiss, reconsider, and take emergency action. On a single day — April 10, 2026, the day after the prior court order — Plaintiffs filed four motions and a response totalling over 40 pages of “coherent legal analysis.” This filing rate was unprecedented in the case’s history and immediately drew the court’s attention.

The AI Issue: AI-Enabled Filing Volume + Fabricated Citations

Judge Abelson identified the “likeliest explanation” for this filing behaviour as AI tool use: the volume, coherence, and speed of the filings — 40 pages in one day — are inconsistent with the limitations typically observed in pro se litigation and strongly suggest AI assistance. This is an important and underexplored dimension of AI in litigation: not just citation hallucination, but AI-enabled volume abuse.

The court did not merely speculate about AI use. It identified specific evidence: in ECF No. 59, Plaintiffs quoted a previous case of the court — “United States v. $17,900 in U.S. Currency” — that does not appear to exist. This is a textbook AI hallucination: a case with a plausible name, plausible citation format, attributed to the right court, but nonexistent.

The court framed the dual problem clearly: AI tools are not inherently improper, but the Aziz plaintiffs’ AI use had produced (1) an avalanche of filings burdening the court, (2) frivolous arguments, (3) misstatements of the record, and (4) fabricated legal authorities. The court cited Maryland precedent holding that pro se litigants are still held to basic accountability standards for their court papers — responsibility that “includes the duty to include accurate factual and legal citations.”

Holdings

1. Four-page filing limit imposed. Going forward, any new motions or oppositions filed by either party are limited to four pages. This restriction applies equally to both sides — an unusual measure designed to reduce the asymmetric burden created by AI-assisted volume filing.
2. AI use itself is not sanctionable, but abuse is. The court explicitly stated that “the use of AI tools for legal research or drafting is not inherently improper.” The sanction trigger is not AI use but its misuse — volume abuse, record misstatements, and fabricated citations.
3. Filing injunction and other sanctions threatened. If Plaintiffs continue to file frivolous motions, include fabricated citations, or violate the page limitation, the court may enjoin them from submitting further filings or impose other sanctions.
4. Show cause order still pending. The order did not resolve the underlying case; Plaintiffs still needed to respond to why their claims against Defendant Aziz should not be dismissed.

“The likeliest explanation is that Plaintiffs are using an artificial intelligence (AI) tool. This inference is further corroborated by numerous hallucinated cases and citations… Parties bear responsibility for the filings they sign, regardless of actual authorship. This responsibility includes the duty to include accurate factual and legal citations.”

— Judge Adam B. Abelson, D. Maryland, April 27, 2026

India Angle: AI-Enabled Litigation Abuse in Indian Courts

Aziz v. USA introduces a dimension of AI risk that is increasingly relevant to Indian courts: not just citation hallucination, but AI-enabled volume filing — the ability to generate coherent, lengthy legal documents at a speed and volume no single pro se litigant could produce manually. This is a structural challenge for Indian courts that already face severe docket pressure.

Relevant Indian Law

• CPC Order VII Rule 11 / Vexatious Litigation: Indian courts can reject plaints that disclose no cause of action and can issue orders restricting vexatious litigants. If AI tools enable the equivalent of the Aziz pattern in Indian courts — dozens of applications filed in rapid succession — courts could invoke Order VII Rule 11 (rejection of plaint), Section 151 CPC (inherent powers), or the Vexatious Litigation (Prevention) Act (available in some states like Maharashtra) to restrict further filings.
• Contempt of Courts Act 1971: Filing documents that misstate the record or include fabricated citations could constitute contempt if the court finds the conduct interferes with the administration of justice. The cumulative effect of multiple such filings may cross into contempt territory even if no single filing does.
• National Court Management System (NCMS) / eCourt Mission Mode Project: Indian courts are digitising filing systems. As e-filing becomes universal, AI-generated volume filing will be technically easier. Courts may need to develop procedural safeguards — page limits, filing frequency caps, or AI disclosure requirements — similar to the four-page limit imposed in Aziz.

Three Practical Tips for Indian Practitioners

1. Volume is not a substitute for merit. The Aziz pattern — filing 40 pages in one day after a court order — is exactly the kind of conduct Indian courts will identify as AI-enabled abuse. If your client wants to challenge a court order, file one well-reasoned application, not seven. Quality, not quantity, is what moves courts.
2. AI-assisted drafting requires proportionate filing discipline. The ease of generating legal documents with AI makes it tempting to file more. Resist this temptation. Every filing you make goes on the court record, incurs costs, and influences the court’s impression of your case. An avalanche of AI-generated motions signals poor judgment, not thoroughness.
3. The “regardless of actual authorship” principle applies in India too. Whether a petition is drafted by a senior advocate, a junior clerk, a paralegal, or an AI tool, the advocate who signs it is personally responsible for its accuracy. This is embedded in the BCI Rules and the Advocates Act. AI does not create a new category of authorship immunity.

Quick Takeaways

• AI-enabled volume filing — not just citation hallucination — is emerging as an independent basis for judicial intervention, including filing restrictions and potential injunctions.
• Courts can and will use the volume of filings, their speed, and their internal coherence as evidence that AI tools were used — even without an admission.
• Filing restrictions (page limits) imposed in response to AI volume abuse can apply equally to both sides — creating unintended procedural disadvantages for the opposing party too.
• The “not inherently improper” framing for AI use is becoming standard: courts are distinguishing between AI as a tool (acceptable) and AI-enabled abuse (sanctionable).
• In India: AI-assisted filing volume abuse is a near-term risk for overburdened courts; the existing procedural toolkit (Order VII Rule 11, Section 151 CPC, vexatious litigant statutes) provides adequate authority to respond.

Deep Dive: Volume Abuse as the Next Frontier of AI Misuse in Litigation

Aziz v. USA represents a qualitative shift in AI litigation risk. The first wave of AI hallucination cases (2023–2025) was primarily about fabricated citations — nonexistent cases cited in briefs. The second wave (2025–2026) added fabricated quotations and mischaracterised holdings. Aziz signals a potential third wave: AI-enabled volume abuse, where the harm is not just incorrect citations but an overwhelming volume of technically coherent filings that burden courts and opposing parties.

Volume abuse is particularly difficult to address because coherent filings, standing alone, are not improper. Each individual motion may raise a legitimate legal question. It is the pattern — the volume, the speed, the cumulative burden — that crosses the line. The court in Aziz addressed this by imposing a page limit rather than dismissing the action or sanctioning specific filings, which is a measured and legally defensible response.

For Indian courts, the volume-abuse risk is more acute because docket management tools are less sophisticated. U.S. federal courts have PACER, electronic filing, automatic notice systems, and individual judge rules that allow relatively rapid identification of a filing pattern. Indian courts — while rapidly digitising — do not yet have the same pattern-detection infrastructure. AI-enabled volume filing could seriously overwhelm already-burdened registry staff and hearing schedules before the court even notices the pattern.

The solution, as the Aziz court recognised, is proportionate procedural restraint — filing limits, frequency caps, and explicit AI disclosure requirements. India’s Supreme Court and High Courts have the inherent power under Article 142 and Section 151 CPC to impose similar restrictions. As AI tools become more accessible to Indian litigants, developing clear and consistent judicial guidance on AI-enabled volume abuse will be as important as developing guidance on citation hallucination.