Eric Andrew Perez v. Dr. Neil C. Evans
U.S. District Court, Southern District of New York · S.D.N.Y. · New York bar guidance
Conduct
Pro se misattributed a 4th Cir. quote to a non-existent SDNY case and used a scrambled docket for a real case holding the opposite.
Consequence
Warning from S.D.N.Y. Judge Broderick; no monetary sanction.
Lesson
Scrambled-docket misattribution (real case, wrong docket) is an AI-hallucination subtype distinct from fabricated citations.
Verified May 14, 2026
- Citation
- Perez v. Evans, No. 1:24-cv-00356 (S.D.N.Y. Sept. 25, 2025) (Broderick, J.)
- Decided
- September 25, 2025
Summary
Pro se plaintiff Eric Andrew Perez filed briefing in a civil rights matter against Dr. Neil C. Evans that the court found contained AI-generated citation defects. Plaintiff attributed a quotation to "Concepcion v. City of New York, No. 05 Civ. 8501 (RJS), 2008 WL 5395720" but no such December 17, 2008 S.D.N.Y. opinion exists; the quoted language actually appears in Armco, Inc. v. Penrod-Stauffer Building Systems, Inc., a Fourth Circuit case. Plaintiff also cited "United States v. Peterson, No. 3:17-cr-00065 (D. Conn. 2018)" for a proposition the cited docket does not support; the docket cited corresponds to United States v. Cook, and the actual Peterson decision (No. 3:18-CR-00049) holds the opposite of plaintiff's asserted rule. Judge Vernon S. Broderick issued a warning, treating both as AI-generated misattributions and scrambled citations.
- AI tool:
- ChatGPT
What sanction did the court impose?
Warning issued by Judge Broderick; no monetary sanction or formal disciplinary referral imposed.
Why does Eric Andrew Perez v. Dr. Neil C. Evans matter for law firms using AI?
Perez v. Evans is a doctrinal step in identifying AI-hallucination subtypes: not just fabricated case citations, but also “scrambled” citations where the docket number, case name, and propositional support do not align with each other. The Concepcion / Armco misattribution and the Peterson / Cook docket swap are both consistent with how generative AI models construct plausible but false citations: real fragments assembled in plausible but unverifiable ways. The court’s warning treats these as functionally equivalent to outright fabrications for sanctions purposes.
For firms cite-checking pro se opposition in S.D.N.Y., the operational lesson is that case-existence checks (does this Westlaw cite resolve to a real opinion) are no longer sufficient. The check must extend to docket-number reconciliation (does this docket number actually correspond to this case name) and quotation provenance (does this quoted language actually appear in the cited case). Those are three separate verification passes; AI hallucinations now span all three.
Implications for your firm
Operational steps a firm reading this case may wish to consider documenting. Strategic and rule-application calls belong to your firm's attorneys.
- Cite-checking pro se opposition in S.D.N.Y. should include a docket-number-to-case-name reconciliation step, not just a case-existence check. The Perez court flagged a real case at a wrong docket.
- Quoted language attributed to one case but actually appearing in a different case is a distinct AI hallucination pattern (the Concepcion / Armco swap). Verify quotation provenance against the cited case, not the case the quotation actually appears in.