Gamez v. County of Fresno: Three-Category AI Defect Pattern Triggers Attorney Show Cause Order | Advocate Prakhar

Case at a Glance

Background

Elio Gamez filed a civil rights and negligence complaint against the County of Fresno in state court; the County removed the action to federal court. When the County filed a motion for a more definite statement, attorney Kevin Little filed an opposition brief on behalf of Gamez. The Magistrate Judge reviewed the brief and identified a three-part pattern of AI hallucination defects, leading to the issuance of a show cause order directing Little to explain the deficiencies by April 20, 2026.

The AI Issue

The court identified three distinct categories of AI-related defects in Little’s brief: first, citations to authority that does not exist; second, apparent quotations from authority that exists but where the quoted text does not appear in the cited opinion; third, material legal representations made without any supporting authority at all. The court’s three-category framework explicitly links the pattern to generative AI reliance: “Such issues suggest that Attorney Little relied on generative artificial intelligence (AI) to draft the opposition brief without ensuring that the generated content was accurate or otherwise supported.”

What the Court Decided

• Attorney Little was ordered to show cause by April 20, 2026 why he should not be sanctioned for the three categories of deficiency in the opposition brief [formal show cause issued].
• Non-existent citations bear the “hallmarks of hallucinated cases created by artificial intelligence tools” — courts have formally adopted this diagnostic language [AI hallucination judicially recognised].
• Fabricated quotations from cases that exist — as in Destfino v. Reiswig, where the quoted language did not appear on the cited page — are separately identified as an AI hallucination pattern [category 2 defect recognised].
• Legal representations without any supporting authority suggest AI-generated content that produces conclusions without the underlying analytical framework [category 3 defect recognised].
• Failure to fully and candidly comply with the show cause order may result in additional sanctions beyond those imposed for the substantive deficiencies [escalating sanction warning].

“Such issues suggest that Attorney Little relied on generative artificial intelligence (AI) to draft the opposition brief without ensuring that the generated content was accurate or otherwise supported.”

— U.S. Magistrate Judge, E.D. California, April 6, 2026

The India Angle

Indian Law Equivalent

The three-category defect taxonomy identified in Gamez maps directly onto three distinct forms of misconduct under Indian professional responsibility law. Category 1 (non-existent citations) engages Rule 14 BCI (no false statements). Category 2 (fabricated quotes from real cases) engages additionally Section 193 BNS 2023 (false evidence). Category 3 (legal representations without supporting authority) engages Rule 14 BCI and the general duty of candour codified under the Advocates Act, 1961. All three categories in combination would likely qualify as “professional misconduct” under Section 35 of the Advocates Act warranting disciplinary action.

Bar Council Rules

BCI Rules 14 (no false statements), 15 (no disrepute), 22 (dignity of court), and 33 (professional conduct) are engaged by all three categories of Gamez-type defects. The three-category framework offers a useful structure for BCI Disciplinary Committee proceedings: committees could assess which categories of AI hallucination defects are present in a reported advocate’s submissions and calibrate the disciplinary response accordingly.

Practical Advice for Indian Advocates

• Run a three-point AI quality check on every brief drafted with AI assistance: (1) verify each citation exists; (2) verify each quoted passage appears verbatim in the cited source; (3) verify each legal proposition is supported by cited authority rather than asserted without attribution.
• When reviewing AI-generated legal representations, be particularly alert to conclusions stated without any cited authority — AI tools often generate confident-sounding legal propositions as if they were self-evident, without the supporting case law that an expert drafter would provide.
• The Eastern District of California’s three-category framework can be used as a self-audit checklist — apply it to any AI-generated brief before signing and filing.

Quick Takeaways

• Three AI defect categories: non-existent cases, false quotes from real cases, unsupported assertions.
• Courts formally diagnose AI hallucination from the three-category pattern in briefs.
• Failing the three-category quality check warrants a show cause order and potential sanctions.

Deep Dive: The Three-Category AI Defect Taxonomy and Its Implications for Brief Review Practice

The Gamez opinion is significant not just for its show cause order against attorney Kevin Little but for the analytical framework it provides to courts and practitioners. By identifying three distinct categories of AI hallucination defects — non-existent citations, fabricated quotes from real cases, and unsupported legal representations — the Eastern District of California magistrate judge has contributed to an emerging taxonomy of AI misconduct in legal filings that is already being adopted (sometimes implicitly, sometimes explicitly) by courts across the US.

Category 1 — the citation to non-existent authority — is the most publicly discussed AI hallucination problem, having been identified in virtually every major AI sanctions case since Mata v. Avianca. Courts have developed efficient procedures for identifying this defect: they run case citations through official databases (Westlaw, Lexis, PACER) and flag any that return no results. In Gamez, the court identified multiple instances where cases cited in the opposition brief simply did not exist, following the same diagnostic process.

Category 2 — fabricated quotations from existing cases — is subtler and potentially more damaging. The cited case exists, can be verified in a database, and may address the same general legal area as the proposition for which it is cited. But the quoted text does not appear in the opinion. AI tools generate these fabricated quotes because they learn to produce text “in the style of” cited sources, producing plausible judicial language that was never actually written by the judge. Detecting Category 2 defects requires reading the actual case, not merely confirming that it exists — a much more time-intensive verification step that courts are now requiring as a baseline.

Category 3 — legal representations without supporting authority — represents the third distinctive AI hallucination pattern. AI tools trained on legal text develop the ability to produce authoritative-sounding legal propositions because they have ingested thousands of legal documents where propositions are stated with confidence. When asked to argue a legal point, the AI produces a confident assertion that may or may not be correct, frequently without the case citations that a human drafter would know to include. In Indian legal practice, where written submissions in writ petitions and appeals must cite authority for every legal proposition, this Category 3 pattern would be particularly visible and immediately problematic.